gen: inline instrumental chooser for all jobs + fix backing-scorer race (#966)

Two fixes for jobs where a backing-vocals song showed **no** inline
chooser and was forced to the heavy `/instrumental` screen — diagnosed
on job `be8231fd` ("The Qemists - Take It Back").

## Diagnosis
The auto-approval verdict is scored by `screens_worker` on lyrics
completion. `mark_audio_complete` is decoupled from status, so when
lyrics finish first and audio is still separating, the backing-vocals
analysis isn't available yet → the backing half is scored `review`
(`analysis_present:false`, `audio_incomplete`). The `audio_worker`
re-score meant to fix this is **skipped once a human opens the editor**
(status `IN_REVIEW`, not `AWAITING_REVIEW`), so the stale "no analysis"
verdict persists forever — even though the analysis later lands and
would score a confident `with_backing` keep.

## Fix 1 — backend: read-time self-correction (race-proof)
`auto_approval_summary` now re-derives the backing verdict from the live
`state_data.backing_vocals_analysis` when the stored verdict was scored
*without* it (`backing_analysis_available` falsy). Pure/read-only —
enforcement & auto-complete are untouched. Unit-tested with the real
job's signals (`corr 0.75`, `-28.7 dB`, `64%` audible → confident
`with_backing`), and that it respects the keep flag and doesn't override
a verdict that already had the analysis.

## Fix 2 — frontend: inline chooser for ALL jobs with both stems
The **"🎬 Your karaoke video will use:"** radio group now shows whenever
both a clean and a with-backing stem exist — not only when the scorer
was confident. Confident → green **"✓ recommended"** badge + confident
copy; not confident → the analysis's recommendation is **preselected**
with a neutral *"Have a listen and choose…"* prompt (no badge). Picking
clean/with-backing **completes inline** ("Complete Track"); **Advanced
mode** is the opt-in to `/instrumental`. The backend `complete` endpoint
already accepts explicit `clean`/`with_backing`, so this is
frontend-only.

**Verified live** on `be8231fd` via the dev server: the chooser now
renders (backing preselected, clean + Advanced offered, CTA "Complete
Track") instead of forcing the advanced screen.

## Tests
- Backend: `auto_approval` suite 121 pass (+3 new self-correction
tests); review/correction-data 142 pass.
- Frontend: full suite **1237 pass**; `ReviewChangesModal` rewritten for
the new prop API incl. the non-confident chooser path. ESLint + tsc
clean.
- i18n: 33 locales, full key parity.

@coderabbitai ignore

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
Andrew Beveridge ☄️andClaude Opus 4.8 authored and GitHub committed 2026-08-30 16:00:33 -04:00
1 parent 7abd0a9617
commit 0bebf5c170
40 files changed
+445 -285

No files matched your search

+33 -1
View File
@@ -79,6 +79,38 @@ def backing_decision(
return {"ok": True, "selection": "clean", "keep_backing": False}
def _backing_verdict_with_late_analysis(job, aa: Dict[str, Any]) -> Dict[str, Any]:
"""Return the stored backing verdict, re-derived from the backing-vocals
analysis when that analysis arrived *after* the verdict was scored.
The auto-approval verdict is scored by ``screens_worker`` on lyrics completion.
When audio separation is still running at that point the backing-vocals analysis
isn't available yet, so the backing half is (correctly) scored ``review``. The
``audio_worker`` re-score is meant to fix this once the analysis lands, but it is
skipped when a human has already opened the editor (the job is ``IN_REVIEW``, not
``AWAITING_REVIEW``) — leaving a stale "no backing analysis" verdict forever.
Whenever the stored verdict was scored without the analysis but the analysis now
exists on the job, re-run the (pure) backing scorer so the instrumental half of
the review reflects reality. Read-only: enforcement/auto-complete is unaffected.
"""
backing = aa.get("backing") or {}
if aa.get("backing_analysis_available"):
return backing
state = getattr(job, "state_data", None) or {}
analysis = state.get("backing_vocals_analysis")
if not analysis:
return backing
from backend.services.auto_approval.scorer import score_backing
fresh = score_backing(analysis)
return {
**backing,
"verdict": fresh.verdict.value,
"non_subjective": fresh.non_subjective,
}
def auto_approval_summary(job, settings) -> Dict[str, Any]:
"""Compact block for the ``/correction-data`` response (C1 per-screen skip).
@@ -87,7 +119,7 @@ def auto_approval_summary(job, settings) -> Dict[str, Any]:
confidently resolved (so it can skip that screen).
"""
aa = (getattr(job, "processing_metadata", None) or {}).get("auto_approval") or {}
backing = aa.get("backing") or {}
backing = _backing_verdict_with_late_analysis(job, aa)
lyrics = aa.get("lyrics") or {}
custom = has_custom_instrumental(job)
@@ -181,6 +181,83 @@ def test_resolve_returns_none_when_with_backing_stem_absent():
assert resolve_auto_instrumental(job, _settings(keep=True)) is None
# ---- late backing analysis self-correction (race fix) -----------------------
def _job_scored_before_analysis(analysis, stems=None):
"""A job whose stored verdict was scored WITHOUT the backing analysis
(screens_worker ran while audio was still separating), but the analysis has
since landed in state_data. Mirrors the real race for job be8231fd."""
if stems is None:
stems = {
"instrumental_clean": "jobs/j/clean.flac",
"instrumental_with_backing": "jobs/j/backing.flac",
}
return SimpleNamespace(
processing_metadata={
"auto_approval": {
# verdict recorded while audio was incomplete: no analysis -> review
"backing": {"verdict": "review", "non_subjective": False},
"lyrics": {"verdict": "review"},
"backing_analysis_available": False,
}
},
state_data={"backing_vocals_analysis": analysis},
backing_preference="auto",
review_mode="auto",
existing_instrumental_gcs_path=None,
file_urls={"stems": stems},
)
# Real signals from be8231fd ("The Qemists - Take It Back"): genuine backing,
# not a misclassified lead (corr 0.75 < 0.80), above the noise floor (-28.7 dB),
# well over the bleed threshold (64% audible) -> confident WITH_BACKING keep.
_GENUINE_BACKING_ANALYSIS = {
"has_audible_content": True,
"audible_percentage": 63.99,
"recommended_selection": "with_backing",
"audible_segments": [],
"stem_comparison": {
"coverage_ratio": 0.9182,
"corr_backing_vocals": 0.7478,
"corr_backing_lead": 0.5107,
"backing_median_db": -28.74,
"backing_audible_fraction": 0.5396,
"lead_audible_fraction": 0.4869,
"lead_overlap_fraction": 0.8305,
"backing_db_std": 5.213,
"flat_fraction": 0.1403,
"error": None,
},
}
def test_summary_self_corrects_when_analysis_arrives_after_scoring():
# Stored verdict is a stale "review" (no analysis); the analysis is now present
# -> the summary re-derives a confident with_backing keep.
job = _job_scored_before_analysis(_GENUINE_BACKING_ANALYSIS)
s = auto_approval_summary(job, _settings(keep=True))
assert s["backing"]["confident"] is True
assert s["backing"]["resolved_selection"] == "with_backing"
def test_summary_self_correction_respects_keep_flag():
# Same late analysis, but the keep flag is off -> still needs a human pick.
job = _job_scored_before_analysis(_GENUINE_BACKING_ANALYSIS)
s = auto_approval_summary(job, _settings(keep=False))
assert s["backing"]["confident"] is False
assert s["backing"]["resolved_selection"] is None
def test_summary_does_not_self_correct_when_analysis_was_present():
# When the stored verdict already had the analysis, trust it verbatim (no
# re-derivation) even if state_data carries an analysis.
job = _job_scored_before_analysis(_GENUINE_BACKING_ANALYSIS)
job.processing_metadata["auto_approval"]["backing_analysis_available"] = True
s = auto_approval_summary(job, _settings(keep=True))
assert s["backing"]["confident"] is False # trusts the stored "review"
# ---- JobCreate coercion (typo must never silently flip intent) ---------------
def test_jobcreate_coerces_out_of_set_preferences():
@@ -0,0 +1,32 @@
# Inline instrumental choice for all jobs + backing-scorer race fix (2026-08-30)
## Problem (job be8231fd "The Qemists - Take It Back")
A clearly-backing-vocals song showed NO inline chooser in the finish modal — just
"Instrumental (clean)" preview + "Proceed to Instrumental Review", forcing the heavy
/instrumental screen.
### Root cause of the missing confident verdict (race)
- `mark_audio_complete` is decoupled from status; screens are gated on lyrics only.
- Lyrics finished first → `screens_worker` scored auto-approval at 17:31 while audio was
still separating → backing analysis absent → verdict `review` (`analysis_present:false`,
`audio_incomplete`). Job parked AWAITING_REVIEW.
- User opened the review (→ IN_REVIEW) before audio finished. When audio finished, the
`audio_worker` re-score (`maybe_auto_complete_review`) bailed at its status gate
(`executor.py` only allows AWAITING_REVIEW for the audio_worker trigger) → the stale,
analysis-less verdict was never refreshed.
- The analysis DID land (state_data.backing_vocals_analysis, recommends with_backing,
corr 0.75, 64% audible) and would score a confident WITH_BACKING keep.
## Fix 1 — read-time self-correction (race-proof)
`auto_approval_summary` (backend/services/auto_approval/instrumental.py): when the stored
backing verdict was scored WITHOUT the analysis (`backing_analysis_available` falsy) but
`job.state_data.backing_vocals_analysis` now exists, re-run `score_backing` on the live
analysis and use that verdict for the summary. Enforcement/auto-complete untouched.
## Fix 2 — inline 3-way chooser for ALL jobs with both stems
Backend `complete` already accepts explicit `clean`/`with_backing` (review.py:958) — so this
is frontend-only. Show the "Your karaoke video will use:" radio group whenever both stems
exist (not only when the scorer was confident). Confident → ✓ recommended badge; not
confident → preselect the analysis's recommended_selection as a neutral "suggested" default.
Picking clean/with_backing completes inline (CTA "Complete Track"); Advanced mode routes to
/instrumental. Submit sends the concrete selection.
@@ -167,10 +167,11 @@ export default function LyricsAnalyzer({
// never shown. The reviewer can opt back in via a "review instrumental anyway"
// toggle in the finish modal (the verdict is confident, not infallible).
const [reviewInstrumentalAnyway, setReviewInstrumentalAnyway] = useState(false)
// Inline single-click override in the finish modal: when the confident backing
// verdict is with_backing, let the reviewer switch to the clean instrumental
// (submits "clean" instead of "auto") without visiting the instrumental screen.
const [useCleanOverride, setUseCleanOverride] = useState(false)
// The reviewer's inline instrumental pick in the finish modal ("clean" |
// "with_backing"), or null to use the recommended default. Completing the review
// submits this concrete selection so the /instrumental screen is only needed for
// the "Advanced mode" opt-in (section muting / custom upload).
const [instrumentalChoice, setInstrumentalChoice] = useState<'clean' | 'with_backing' | null>(null)
const [vocalsAudioUrl, setVocalsAudioUrl] = useState<string | null>(null)
@@ -297,6 +298,35 @@ export default function LyricsAnalyzer({
!hasExistingInstrumental &&
Boolean(data.auto_approval?.backing?.confident)
// Inline instrumental choice: offer the clean-vs-with-backing picker in the finish
// modal whenever BOTH stems exist — even when the auto-scorer wasn't confident — so
// the reviewer never has to visit the heavy /instrumental screen just to make the
// basic choice. That screen becomes the "Advanced mode" opt-in.
const instrumentalStemOptions = data.instrumental_options ?? []
const hasCleanStem = instrumentalStemOptions.some((o) => o.id === 'clean' && o.audio_url)
const hasBackingStem = instrumentalStemOptions.some((o) => o.id === 'with_backing' && o.audio_url)
const offerInlineInstrumentalChoice =
!isReadOnly && !isLocalMode && Boolean(jobId) && !hasExistingInstrumental && hasCleanStem && hasBackingStem
// Which instrumental to preselect / recommend: the confident server resolution,
// else the backing-vocals analysis's recommendation, else keep backing where a
// backing stem exists ("retain where possible"), falling back to clean. Both
// sources can carry non-stem values (null / "review_needed"), so validate.
const asStem = (v: unknown): 'clean' | 'with_backing' | undefined =>
v === 'clean' || v === 'with_backing' ? v : undefined
const rawRecommended =
asStem(data.auto_approval?.backing?.resolved_selection) ??
asStem(data.backing_vocals_analysis?.recommended_selection) ??
(hasBackingStem ? 'with_backing' : 'clean')
// Clamp to a stem that actually exists.
const recommendedInstrumental: 'clean' | 'with_backing' =
rawRecommended === 'with_backing' && !hasBackingStem
? 'clean'
: rawRecommended === 'clean' && !hasCleanStem && hasBackingStem
? 'with_backing'
: rawRecommended
const currentInstrumental: 'clean' | 'with_backing' = instrumentalChoice ?? recommendedInstrumental
// Close the run modal once suggestions arrive (panel takes over)
useEffect(() => {
if (autoCorrect.status === 'reviewing') {
@@ -1096,16 +1126,18 @@ export default function LyricsAnalyzer({
toast.error('Failed to start video generation. Please try again.')
setIsSubmitting(false) // Keep modal open so the user can retry
}
} else if (autoInstrumentalConfident && !reviewInstrumentalAnyway && jobId) {
// Per-screen skip (C1): the backing decision was confidently auto-resolved
// -> complete the review now and skip the instrumental screen entirely.
// "auto" lets the backend resolve the stored verdict; the inline override
// sends "clean" explicitly when the reviewer switched away from backing.
const selection = useCleanOverride ? 'clean' : 'auto'
} else if ((offerInlineInstrumentalChoice || autoInstrumentalConfident) && !reviewInstrumentalAnyway && jobId) {
// Inline instrumental choice: complete the review now with the reviewer's
// pick (or the recommended default) and skip the /instrumental screen.
// When both stems exist we submit the concrete selection; the confident
// single-stem case falls back to "auto" (backend resolves the verdict).
const selection = offerInlineInstrumentalChoice ? currentInstrumental : 'auto'
toast.success(
useCleanOverride
selection === 'clean'
? 'Lyrics saved! Using clean instrumental, generating video...'
: 'Lyrics saved! Instrumental auto-selected, generating video...'
: selection === 'with_backing'
? 'Lyrics saved! Using instrumental with backing vocals, generating video...'
: 'Lyrics saved! Instrumental auto-selected, generating video...'
)
try {
const correctionData = await lyricsReviewApi.getCorrectionData(jobId)
@@ -1113,9 +1145,9 @@ export default function LyricsAnalyzer({
setShowSuccess(true)
setCountdown(3)
} catch (err) {
// Fall back to the manual instrumental screen if auto-resolution fails
// (e.g. the verdict wasn't resolvable server-side) — never strand the user.
console.error('Auto instrumental selection failed, falling back to review:', err)
// Fall back to the manual instrumental screen if completion fails
// (e.g. the selected stem is unexpectedly missing) — never strand the user.
console.error('Inline instrumental selection failed, falling back to review:', err)
setIsReviewModalOpen(false)
setIsSubmitting(false)
toast('Please choose your instrumental to finish.')
@@ -1145,7 +1177,7 @@ export default function LyricsAnalyzer({
toast.error('Failed to submit corrections. Please try again.')
setIsSubmitting(false) // Reset on error so user can retry
}
}, [apiClient, data, timingOffsetMs, editLog, isLocalMode, jobId, hasExistingInstrumental, isDuet, autoInstrumentalConfident, reviewInstrumentalAnyway, useCleanOverride])
}, [apiClient, data, timingOffsetMs, editLog, isLocalMode, jobId, hasExistingInstrumental, isDuet, autoInstrumentalConfident, offerInlineInstrumentalChoice, currentInstrumental, reviewInstrumentalAnyway])
// Play segment handler
const handlePlaySegment = useCallback(
@@ -1693,13 +1725,14 @@ export default function LyricsAnalyzer({
apiClient={apiClient}
timingOffsetMs={timingOffsetMs}
isDuet={isDuet}
completesReview={hasExistingInstrumental || (autoInstrumentalConfident && !reviewInstrumentalAnyway)}
autoInstrumentalConfident={autoInstrumentalConfident}
autoInstrumentalSelection={data.auto_approval?.backing?.resolved_selection ?? null}
completesReview={hasExistingInstrumental || ((offerInlineInstrumentalChoice || autoInstrumentalConfident) && !reviewInstrumentalAnyway)}
offerInlineChoice={offerInlineInstrumentalChoice}
autoConfident={autoInstrumentalConfident}
recommendedSelection={recommendedInstrumental}
currentSelection={currentInstrumental}
onSelectInstrumental={setInstrumentalChoice}
reviewInstrumentalAnyway={reviewInstrumentalAnyway}
onToggleReviewInstrumental={setReviewInstrumentalAnyway}
cleanOverride={useCleanOverride}
onInstrumentalChoiceChange={(choice) => setUseCleanOverride(choice === 'clean')}
/>
<AddLyricsModal
@@ -127,50 +127,47 @@ describe('ReviewChangesModal', () => {
expect(screen.getByText(/manual corrections detected/i)).toBeInTheDocument()
})
it('renders the backing-vocals waveform when backing is kept and waveform data is available', () => {
const bothStems = [
{ id: 'clean', label: 'Clean', audio_url: 'http://x/clean.ogg' },
{ id: 'with_backing', label: 'Backing', audio_url: 'http://x/backing.ogg' },
]
const waveformApiClient = () => ({
generatePreviewVideo: jest.fn(),
getPreviewVideoStatus: jest.fn(),
getPreviewVideoUrl: jest.fn(),
getWaveformData: jest.fn().mockResolvedValue({ amplitudes: [0.1, 0.2], duration_seconds: 10 }),
})
it('renders the backing-vocals waveform when backing is the current selection', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
const apiClient = {
generatePreviewVideo: jest.fn(),
getPreviewVideoStatus: jest.fn(),
getPreviewVideoUrl: jest.fn(),
getWaveformData: jest.fn().mockResolvedValue({ amplitudes: [0.1, 0.2], duration_seconds: 10 }),
}
const apiClient = waveformApiClient()
render(
<ReviewChangesModal
{...defaultProps}
apiClient={apiClient as any}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="with_backing"
data={makeData({
corrected_segments: segments as any,
instrumental_options: [
{ id: 'clean', label: 'Clean', audio_url: 'http://x/clean.ogg' },
{ id: 'with_backing', label: 'Backing', audio_url: 'http://x/backing.ogg' },
] as any,
})}
offerInlineChoice
autoConfident
recommendedSelection="with_backing"
currentSelection="with_backing"
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
// Hint text from BackingVocalsWaveform
expect(screen.getByText(/click to hear this part/i)).toBeInTheDocument()
expect(apiClient.getWaveformData).toHaveBeenCalled()
})
it('does not render the backing-vocals waveform when the backing stem URL is unavailable', () => {
it('does not render the backing-vocals waveform when the backing stem is unavailable', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
const apiClient = {
generatePreviewVideo: jest.fn(),
getPreviewVideoStatus: jest.fn(),
getPreviewVideoUrl: jest.fn(),
getWaveformData: jest.fn().mockResolvedValue({ amplitudes: [0.1], duration_seconds: 10 }),
}
const apiClient = waveformApiClient()
render(
<ReviewChangesModal
{...defaultProps}
apiClient={apiClient as any}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="with_backing"
autoConfident
recommendedSelection="clean"
currentSelection="clean"
data={makeData({
corrected_segments: segments as any,
instrumental_options: [{ id: 'clean', label: 'Clean', audio_url: 'http://x/clean.ogg' }] as any,
@@ -181,33 +178,27 @@ describe('ReviewChangesModal', () => {
expect(apiClient.getWaveformData).not.toHaveBeenCalled()
})
it('does not render the backing-vocals waveform when the clean instrumental was chosen', () => {
it('does not render the backing-vocals waveform when clean is the current selection', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
const apiClient = {
generatePreviewVideo: jest.fn(),
getPreviewVideoStatus: jest.fn(),
getPreviewVideoUrl: jest.fn(),
getWaveformData: jest.fn().mockResolvedValue({ amplitudes: [0.1], duration_seconds: 10 }),
}
const apiClient = waveformApiClient()
render(
<ReviewChangesModal
{...defaultProps}
apiClient={apiClient as any}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="clean"
data={makeData({ corrected_segments: segments as any })}
offerInlineChoice
currentSelection="clean"
recommendedSelection="with_backing"
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
expect(screen.queryByText(/click to hear this part/i)).not.toBeInTheDocument()
expect(apiClient.getWaveformData).not.toHaveBeenCalled()
})
it('calls onSubmit when button clicked with valid segments', async () => {
const user = userEvent.setup()
const onSubmit = jest.fn()
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
render(
<ReviewChangesModal
{...defaultProps}
@@ -215,7 +206,6 @@ describe('ReviewChangesModal', () => {
data={makeData({ corrected_segments: segments as any })}
/>
)
await user.click(screen.getByRole('button', { name: /proceed to instrumental/i }))
expect(onSubmit).toHaveBeenCalledTimes(1)
})
@@ -229,85 +219,66 @@ describe('ReviewChangesModal', () => {
data={makeData({ corrected_segments: segments as any })}
/>
)
expect(screen.getByRole('button', { name: /complete track/i })).toBeInTheDocument()
expect(
screen.queryByRole('button', { name: /proceed to instrumental/i })
).not.toBeInTheDocument()
expect(screen.queryByRole('button', { name: /proceed to instrumental/i })).not.toBeInTheDocument()
})
// Per-screen skip (C1): backing decision auto-resolved. The final-output choice
// is a single radio group ("Your karaoke video will use:") whose recommended
// option is the auto-selected instrumental; "Advanced mode" is the escape hatch.
it('shows the final-output radio group + backing note when backing is confident', () => {
it('shows the chooser with a green "recommended" badge + backing note when confident', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
render(
<ReviewChangesModal
{...defaultProps}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="with_backing"
data={makeData({ corrected_segments: segments as any })}
offerInlineChoice
autoConfident
recommendedSelection="with_backing"
currentSelection="with_backing"
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
expect(screen.getByRole('button', { name: /complete track/i })).toBeInTheDocument()
expect(screen.getByText(/your karaoke video will use/i)).toBeInTheDocument()
// Recommended option = backing; escape hatch = Advanced mode.
const backingRadio = screen.getByRole('radio', { name: /instrumental \+ backing vocals/i })
expect(backingRadio).toBeChecked()
expect(screen.getByRole('radio', { name: /instrumental \+ backing vocals/i })).toBeChecked()
expect(screen.getByText(/recommended/i)).toBeInTheDocument()
expect(screen.queryByText(/^suggested$/i)).not.toBeInTheDocument()
expect(screen.getByText(/backing vocals in the instrumental/i)).toBeInTheDocument()
expect(
screen.getByRole('radio', { name: /advanced mode \(edit backing vocals or upload your own instrumental\)/i })
).toBeInTheDocument()
})
it('offers a Clean instrumental radio only when a clean stem exists alongside backing', () => {
it('shows the chooser with a preselected default and no badge when NOT confident', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
render(
<ReviewChangesModal
{...defaultProps}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="with_backing"
data={makeData({
corrected_segments: segments as any,
instrumental_options: [
{ id: 'clean', label: 'Clean', audio_url: 'http://x/clean.ogg' },
{ id: 'with_backing', label: 'Backing', audio_url: 'http://x/backing.ogg' },
] as any,
})}
offerInlineChoice
recommendedSelection="with_backing"
currentSelection="with_backing"
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
expect(screen.getByRole('radio', { name: /instrumental \+ backing vocals/i })).toBeChecked()
expect(screen.getByRole('radio', { name: /^clean instrumental$/i })).toBeInTheDocument()
expect(screen.getByRole('radio', { name: /advanced mode/i })).toBeInTheDocument()
// No "recommended" badge when the scorer wasn't confident — just the neutral
// prompt and a preselected default.
expect(screen.queryByText(/recommended/i)).not.toBeInTheDocument()
expect(screen.getByText(/have a listen and choose/i)).toBeInTheDocument()
})
it('reflects the clean choice in the note when the reviewer selects the clean radio', () => {
// The decision state is owned by LyricsAnalyzer and passed via cleanOverride;
// selecting clean marks that radio and swaps the note.
it('checks the clean radio when clean is the current selection', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
const options = [
{ id: 'clean', label: 'Clean', audio_url: 'http://x/clean.ogg' },
{ id: 'with_backing', label: 'Backing', audio_url: 'http://x/backing.ogg' },
]
const { rerender } = render(
render(
<ReviewChangesModal
{...defaultProps}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="with_backing"
data={makeData({ corrected_segments: segments as any, instrumental_options: options as any })}
/>
)
expect(screen.getByText(/backing vocals in the instrumental/i)).toBeInTheDocument()
rerender(
<ReviewChangesModal
{...defaultProps}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="with_backing"
cleanOverride
data={makeData({ corrected_segments: segments as any, instrumental_options: options as any })}
offerInlineChoice
autoConfident
recommendedSelection="with_backing"
currentSelection="clean"
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
expect(screen.getByRole('radio', { name: /^clean instrumental$/i })).toBeChecked()
@@ -316,65 +287,61 @@ describe('ReviewChangesModal', () => {
it('selecting the clean radio reports the choice and clears the review-anyway flag', async () => {
const user = userEvent.setup()
const onInstrumentalChoiceChange = jest.fn()
const onSelectInstrumental = jest.fn()
const onToggleReviewInstrumental = jest.fn()
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
render(
<ReviewChangesModal
{...defaultProps}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="with_backing"
onInstrumentalChoiceChange={onInstrumentalChoiceChange}
offerInlineChoice
recommendedSelection="with_backing"
currentSelection="with_backing"
onSelectInstrumental={onSelectInstrumental}
onToggleReviewInstrumental={onToggleReviewInstrumental}
data={makeData({
corrected_segments: segments as any,
instrumental_options: [
{ id: 'clean', label: 'Clean', audio_url: 'http://x/clean.ogg' },
{ id: 'with_backing', label: 'Backing', audio_url: 'http://x/backing.ogg' },
] as any,
})}
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
await user.click(screen.getByRole('radio', { name: /^clean instrumental$/i }))
expect(onInstrumentalChoiceChange).toHaveBeenCalledWith('clean')
expect(onSelectInstrumental).toHaveBeenCalledWith('clean')
expect(onToggleReviewInstrumental).toHaveBeenCalledWith(false)
})
it('uses the clean-case Advanced-mode wording when clean is already the default', () => {
it('selecting the backing radio reports with_backing', async () => {
const user = userEvent.setup()
const onSelectInstrumental = jest.fn()
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
render(
<ReviewChangesModal
{...defaultProps}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="clean"
data={makeData({
corrected_segments: segments as any,
instrumental_options: [
{ id: 'clean', label: 'Clean', audio_url: 'http://x/clean.ogg' },
] as any,
})}
offerInlineChoice
recommendedSelection="clean"
currentSelection="clean"
onSelectInstrumental={onSelectInstrumental}
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
await user.click(screen.getByRole('radio', { name: /instrumental \+ backing vocals/i }))
expect(onSelectInstrumental).toHaveBeenCalledWith('with_backing')
})
it('uses the clean-context Advanced-mode wording when clean is recommended', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
render(
<ReviewChangesModal
{...defaultProps}
completesReview
offerInlineChoice
autoConfident
recommendedSelection="clean"
currentSelection="clean"
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
expect(
screen.getByRole('radio', { name: /advanced mode \(review or upload a custom instrumental\)/i })
).toBeInTheDocument()
// No clean-override radio when clean is already the recommended default.
expect(screen.queryByRole('radio', { name: /^clean instrumental$/i })).not.toBeInTheDocument()
})
it('shows the clean-instrumental note when the resolved selection is clean', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
render(
<ReviewChangesModal
{...defaultProps}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="clean"
data={makeData({ corrected_segments: segments as any })}
/>
)
expect(screen.getByText(/we'll use the clean instrumental/i)).toBeInTheDocument()
})
@@ -386,17 +353,18 @@ describe('ReviewChangesModal', () => {
<ReviewChangesModal
{...defaultProps}
completesReview
autoInstrumentalConfident
autoInstrumentalSelection="with_backing"
offerInlineChoice
recommendedSelection="with_backing"
currentSelection="with_backing"
onToggleReviewInstrumental={onToggleReviewInstrumental}
data={makeData({ corrected_segments: segments as any })}
data={makeData({ corrected_segments: segments as any, instrumental_options: bothStems as any })}
/>
)
await user.click(screen.getByRole('radio', { name: /advanced mode/i }))
expect(onToggleReviewInstrumental).toHaveBeenCalledWith(true)
})
it('hides the final-output radio group when backing is not confident', () => {
it('hides the chooser when neither confident nor both stems are available', () => {
const segments = [{ text: 'Hello', words: [], start_time: 0, end_time: 1 }]
render(
<ReviewChangesModal
@@ -41,21 +41,23 @@ interface ReviewChangesModalProps {
/** When true (e.g. tenant / uploaded-instrumental jobs) approving here completes the
* track directly — there is no instrumental-review step — so the CTA reflects that. */
completesReview?: boolean
/** Per-screen skip (C1): the backing decision was confidently auto-resolved, so
* approving here can complete the whole job and the instrumental screen is skipped.
* Shows a "review instrumental anyway" escape hatch. */
autoInstrumentalConfident?: boolean
/** The server-resolved instrumental ("clean" | "with_backing" | ...) for the note. */
autoInstrumentalSelection?: string | null
/** Whether the reviewer opted back into the instrumental screen. */
/** Show the inline instrumental chooser ("Your karaoke video will use:") — true
* whenever both a clean and a with-backing stem exist, so the reviewer picks the
* instrumental here and the heavy /instrumental screen becomes an opt-in ("Advanced
* mode"). Independent of scorer confidence. */
offerInlineChoice?: boolean
/** The auto-scorer was confident about the backing decision → the recommended
* option gets a green "✓ recommended" badge + the confident helper copy. When
* false the recommended option is a softer "suggested" default. */
autoConfident?: boolean
/** Which instrumental to badge + preselect ("clean" | "with_backing"). */
recommendedSelection?: 'clean' | 'with_backing' | null
/** The instrumental currently chosen for the produced video. */
currentSelection?: 'clean' | 'with_backing'
onSelectInstrumental?: (choice: 'clean' | 'with_backing') => void
/** Whether the reviewer opted into the full instrumental screen ("Advanced mode"). */
reviewInstrumentalAnyway?: boolean
onToggleReviewInstrumental?: (val: boolean) => void
/** Inline single-click override, driven by the "Audio:" toggle's clean pill:
* true once the reviewer has switched the auto-selected backing instrumental to
* the clean one. Only offered when the verdict is with_backing and a clean stem
* exists (no need to show it when clean is already the default). */
cleanOverride?: boolean
onInstrumentalChoiceChange?: (choice: 'clean' | 'with_backing') => void
}
export default function ReviewChangesModal({
@@ -68,12 +70,13 @@ export default function ReviewChangesModal({
timingOffsetMs = 0,
isDuet,
completesReview = false,
autoInstrumentalConfident = false,
autoInstrumentalSelection = null,
offerInlineChoice = false,
autoConfident = false,
recommendedSelection = null,
currentSelection,
onSelectInstrumental,
reviewInstrumentalAnyway = false,
onToggleReviewInstrumental,
cleanOverride = false,
onInstrumentalChoiceChange,
}: ReviewChangesModalProps) {
const t = useTranslations('lyricsReview.modals.reviewChanges')
// Instrumental variant labels are shared with the preview toggle.
@@ -89,60 +92,46 @@ export default function ReviewChangesModal({
const [previewTime, setPreviewTime] = useState(0)
const instrumentalOptions = data.instrumental_options
const getWaveformData = apiClient?.getWaveformData
// Only offer the backing waveform (which clicks through to the backing stem)
// when that stem is actually playable — otherwise the seek would switch audio
// to the wrong/absent track.
const hasBackingStem = !!instrumentalOptions?.some(
(o) => o.id === 'with_backing' && o.audio_url
)
const hasCleanStem = !!instrumentalOptions?.some(
(o) => o.id === 'clean' && o.audio_url
)
// Render the "Your karaoke video will use:" chooser when both stems exist (the
// reviewer picks inline) or when a confident single-stem verdict still wants the
// recommended-option + Advanced-mode framing.
const showChoiceBlock = offerInlineChoice || autoConfident
// The instrumental currently chosen for the produced video (ignoring Advanced).
const selected: 'clean' | 'with_backing' =
currentSelection ?? recommendedSelection ?? (hasBackingStem ? 'with_backing' : 'clean')
// Only offer the backing waveform (which clicks through to the backing stem)
// when that stem is playable and the reviewer is currently keeping backing.
const showBackingWaveform =
autoInstrumentalConfident &&
autoInstrumentalSelection === 'with_backing' &&
hasBackingStem &&
!!getWaveformData
!reviewInstrumentalAnyway && selected === 'with_backing' && hasBackingStem && !!getWaveformData
// Split the audio toggle into "Instrumental + backing vocals" / "Clean
// instrumental" pills only when the auto-selection is the backing instrumental
// AND a clean stem exists to switch to (i.e. the two are meaningfully
// different). When clean is already the default there is nothing to offer.
const offerInstrumentalChoice =
autoInstrumentalConfident &&
autoInstrumentalSelection === 'with_backing' &&
hasCleanStem &&
hasBackingStem
const cleanChosen = offerInstrumentalChoice && cleanOverride
// The three mutually-exclusive final-output outcomes, surfaced as one radio
// group so it's unambiguous which instrumental the produced video will use.
// - 'auto' → the auto-selected instrumental (backing, or clean when that's
// the verdict). Submits "auto"/backing.
// - 'clean' → override to the clean instrumental (only offered when backing
// was auto-picked and a clean stem exists). Submits "clean".
// - 'advanced' → opt into the full instrumental screen (mute sections / upload).
const decision: 'auto' | 'clean' | 'advanced' = reviewInstrumentalAnyway
? 'advanced'
: cleanChosen
? 'clean'
: 'auto'
const selectAuto = () => {
const chooseInstrumental = (choice: 'clean' | 'with_backing') => {
onToggleReviewInstrumental?.(false)
if (autoInstrumentalSelection === 'with_backing') {
onInstrumentalChoiceChange?.('with_backing')
previewRef.current?.auditionInstrumental('with_backing')
} else {
previewRef.current?.auditionInstrumental('clean')
}
onSelectInstrumental?.(choice)
previewRef.current?.auditionInstrumental(choice)
}
const selectClean = () => {
onToggleReviewInstrumental?.(false)
onInstrumentalChoiceChange?.('clean')
previewRef.current?.auditionInstrumental('clean')
const chooseAdvanced = () => onToggleReviewInstrumental?.(true)
// A firm "✓ recommended" badge next to the recommended option — only when the
// scorer was confident. When it wasn't, the option is still preselected but
// carries no badge (the neutral "have a listen and choose" prompt says it's the
// reviewer's call), avoiding a subtle recommended-vs-suggested distinction that
// doesn't survive translation.
const recommendedBadge = (value: 'clean' | 'with_backing') => {
if (!autoConfident || recommendedSelection !== value) return null
return (
<span className="text-xs text-green-600 dark:text-green-400 whitespace-nowrap">
✓ {t('recommended')}
</span>
)
}
const selectAdvanced = () => onToggleReviewInstrumental?.(true)
const handleSubmit = () => {
if (isSubmitting) return
@@ -165,7 +154,7 @@ export default function ReviewChangesModal({
timingOffsetMs={timingOffsetMs}
isDuet={isDuet}
instrumentalOptions={instrumentalOptions}
autoSelection={autoInstrumentalSelection}
autoSelection={selected}
onTimeUpdate={setPreviewTime}
/>
@@ -192,45 +181,41 @@ export default function ReviewChangesModal({
</div>
)}
{/* Per-screen skip (C1): the backing decision was auto-resolved, so this
modal completes the whole job. The single radio group makes the
final-output choice explicit (and the escape hatch is just another
option — "Advanced mode"), separate from the preview-audio toggle
above which only controls what you hear right now. */}
{autoInstrumentalConfident && (
{/* "Your karaoke video will use:" — the explicit final-output choice,
separate from the preview-audio toggle above (which only controls what
you hear). Shown whenever both stems exist so the reviewer picks the
instrumental here; the full /instrumental screen is the "Advanced mode"
opt-in. When the scorer was confident the recommended option carries a
green ✓; otherwise it's a softer preselected "suggested" default. */}
{showChoiceBlock && (
<div className="rounded-lg border border-purple-500/40 bg-purple-500/5 p-3 text-sm space-y-3">
<p className="font-medium">🎬 {t('finalChoiceLabel')}</p>
<div role="radiogroup" className="space-y-2">
{/* Recommended: the auto-selected instrumental */}
<label className="flex items-center gap-2 cursor-pointer">
<input
type="radio"
name="final-instrumental"
className="h-4 w-4"
checked={decision === 'auto'}
onChange={selectAuto}
/>
<span>
{autoInstrumentalSelection === 'with_backing'
? tPreview('audioInstrumentalBacking')
: tPreview('audioInstrumentalClean')}
</span>
<span className="text-xs text-green-600 dark:text-green-400 whitespace-nowrap">
✓ {t('recommended')}
</span>
</label>
{/* Clean override — only when backing was auto-picked and a clean stem exists */}
{offerInstrumentalChoice && (
{hasBackingStem && (
<label className="flex items-center gap-2 cursor-pointer">
<input
type="radio"
name="final-instrumental"
className="h-4 w-4"
checked={decision === 'clean'}
onChange={selectClean}
checked={!reviewInstrumentalAnyway && selected === 'with_backing'}
onChange={() => chooseInstrumental('with_backing')}
/>
<span>{tPreview('audioInstrumentalBacking')}</span>
{recommendedBadge('with_backing')}
</label>
)}
{hasCleanStem && (
<label className="flex items-center gap-2 cursor-pointer">
<input
type="radio"
name="final-instrumental"
className="h-4 w-4"
checked={!reviewInstrumentalAnyway && selected === 'clean'}
onChange={() => chooseInstrumental('clean')}
/>
<span>{tPreview('audioInstrumentalClean')}</span>
{recommendedBadge('clean')}
</label>
)}
@@ -240,21 +225,27 @@ export default function ReviewChangesModal({
type="radio"
name="final-instrumental"
className="h-4 w-4"
checked={decision === 'advanced'}
onChange={selectAdvanced}
checked={!!reviewInstrumentalAnyway}
onChange={chooseAdvanced}
/>
<span>
{autoInstrumentalSelection === 'with_backing'
? t('advancedMode')
: t('advancedModeClean')}
{recommendedSelection === 'clean' ? t('advancedModeClean') : t('advancedMode')}
</span>
</label>
</div>
{/* Context for the current decision */}
{decision === 'auto' && autoInstrumentalSelection === 'with_backing' && (
{/* Context for the current choice */}
{!reviewInstrumentalAnyway && (
<>
<p className="text-muted-foreground">{t('autoInstrumentalBacking')}</p>
{!autoConfident ? (
<p className="text-muted-foreground">{t('chooseInstrumentalPrompt')}</p>
) : selected === 'with_backing' ? (
<p className="text-muted-foreground">{t('autoInstrumentalBacking')}</p>
) : recommendedSelection === 'clean' ? (
<p className="text-muted-foreground">{t('autoInstrumentalClean')}</p>
) : (
<p className="text-muted-foreground">{t('autoInstrumentalCleanChosen')}</p>
)}
{showBackingWaveform && getWaveformData && (
<BackingVocalsWaveform
getWaveformData={getWaveformData}
@@ -264,12 +255,6 @@ export default function ReviewChangesModal({
)}
</>
)}
{decision === 'auto' && autoInstrumentalSelection !== 'with_backing' && (
<p className="text-muted-foreground">{t('autoInstrumentalClean')}</p>
)}
{decision === 'clean' && (
<p className="text-muted-foreground">{t('autoInstrumentalCleanChosen')}</p>
)}
</div>
)}
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "سيستخدم فيديو الكاريوكي الخاص بك:",
"recommended": "موصى به",
"autoInstrumentalBacking": "سنحتفظ بالأصوات الخلفية في النسخة الموسيقية. استمع إليها — إذا تمكنت من سماع صوت المغني الرئيسي، بدّل إلى \"موسيقى صافية\" أعلاه.",
"advancedMode": "الوضع المتقدم (تعديل الأصوات الخلفية أو رفع الموسيقى الخاصة بك)"
"advancedMode": "الوضع المتقدم (تعديل الأصوات الخلفية أو رفع الموسيقى الخاصة بك)",
"chooseInstrumentalPrompt": "استمع واختر الموسيقى التي تود استخدامها في الفيديو الخاص بك."
},
"searchLyrics": {
"artist": "الفنان",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "El teu vídeo de karaoke utilitzarà:",
"recommended": "recomanat",
"autoInstrumentalBacking": "Mantenim els cors a l'instrumental. Escolta-ho — si sents la veu del cantant principal, canvia a «Instrumental net» més amunt.",
"advancedMode": "Mode avançat (edita les veus de fons o puja la teva pròpia instrumental)"
"advancedMode": "Mode avançat (edita les veus de fons o puja la teva pròpia instrumental)",
"chooseInstrumentalPrompt": "Escolta-ho i tria quina instrumental vols fer servir al teu vídeo."
},
"searchLyrics": {
"artist": "Artista",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Ve tvém karaoke videu se použije:",
"recommended": "doporučeno",
"autoInstrumentalBacking": "V instrumentální verzi ponecháváme doprovodné vokály. Poslechni si to — pokud v ní slyšíš hlas hlavního zpěváka, přepni výše na možnost „Čistá instrumentální verze“.",
"advancedMode": "Pokročilý režim (uprav doprovodné vokály nebo nahraj vlastní instrumentální verzi)"
"advancedMode": "Pokročilý režim (uprav doprovodné vokály nebo nahraj vlastní instrumentální verzi)",
"chooseInstrumentalPrompt": "Poslechni si to a vyber, kterou instrumentální verzi má tvé video použít."
},
"searchLyrics": {
"artist": "Interpret",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Din karaokevideo vil bruge:",
"recommended": "anbefales",
"autoInstrumentalBacking": "Vi beholder baggrundsvokalerne i instrumentalen. Prøv at lytte — hvis du kan høre forsangerens stemme trænge igennem, så skift til “Ren instrumental” ovenfor.",
"advancedMode": "Avanceret tilstand (rediger baggrundsvokaler eller upload dit eget instrumentalspor)"
"advancedMode": "Avanceret tilstand (rediger baggrundsvokaler eller upload dit eget instrumentalspor)",
"chooseInstrumentalPrompt": "Prøv at lytte, og vælg hvilket instrumentalspor din video skal bruge."
},
"searchLyrics": {
"artist": "Kunstner",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Dein Karaoke-Video verwendet:",
"recommended": "empfohlen",
"autoInstrumentalBacking": "Wir behalten den Hintergrundgesang im Instrumental. Hör mal rein — wenn du die Stimme des Hauptsängers heraushören kannst, wechsle oben zu „reines Instrumental“.",
"advancedMode": "Erweiterter Modus (Hintergrundgesang bearbeiten oder dein eigenes Instrumental hochladen)"
"advancedMode": "Erweiterter Modus (Hintergrundgesang bearbeiten oder dein eigenes Instrumental hochladen)",
"chooseInstrumentalPrompt": "Hör rein und wähle, welches Instrumental dein Video verwenden soll."
},
"searchLyrics": {
"artist": "Künstler",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Το βίντεο καραόκε σου θα χρησιμοποιήσει:",
"recommended": "προτείνεται",
"autoInstrumentalBacking": "Κρατάμε τα δεύτερα φωνητικά στο ορχηστρικό. Άκουσέ το — αν ακούγεται η φωνή του βασικού τραγουδιστή, άλλαξε σε «Καθαρό ορχηστρικό» παραπάνω.",
"advancedMode": "Λειτουργία για προχωρημένους (επεξεργάσου τα δεύτερα φωνητικά ή ανέβασε το δικό σου ορχηστρικό)"
"advancedMode": "Λειτουργία για προχωρημένους (επεξεργάσου τα δεύτερα φωνητικά ή ανέβασε το δικό σου ορχηστρικό)",
"chooseInstrumentalPrompt": "Άκουσε και διάλεξε ποιο ορχηστρικό θα χρησιμοποιηθεί στο βίντεό σου."
},
"searchLyrics": {
"artist": "Καλλιτέχνης",
+1
View File
@@ -1241,6 +1241,7 @@
"autoInstrumentalCleanChosen": "You've chosen the clean instrumental (no backing vocals) for this track.",
"saving": "Saving...",
"finalChoiceLabel": "Your karaoke video will use:",
"chooseInstrumentalPrompt": "Have a listen and choose which instrumental your video should use.",
"recommended": "recommended",
"advancedMode": "Advanced mode (edit backing vocals or upload your own instrumental)",
"advancedModeClean": "Advanced mode (review or upload a custom instrumental)"
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Tu video de karaoke usará:",
"recommended": "recomendado",
"autoInstrumentalBacking": "Hemos dejado los coros en el instrumental. Escúchalo y, si notas que se cuela la voz del cantante principal, cambia a “Instrumental limpio” arriba.",
"advancedMode": "Modo avanzado (edita los coros o sube tu propio instrumental)"
"advancedMode": "Modo avanzado (edita los coros o sube tu propio instrumental)",
"chooseInstrumentalPrompt": "Escucha y elige qué instrumental debe usar tu video."
},
"searchLyrics": {
"artist": "Artista",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Karaokevideossasi käytetään:",
"recommended": "suositeltu",
"autoInstrumentalBacking": "Säilytämme taustalaulut instrumentaalissa. Kuuntele — jos päälaulajan ääni kuuluu läpi, vaihda yllä olevaan ”Puhdas instrumentaali” -vaihtoehtoon.",
"advancedMode": "Edistynyt tila (muokkaa taustalauluja tai lataa oma instrumentaaliversiosi)"
"advancedMode": "Edistynyt tila (muokkaa taustalauluja tai lataa oma instrumentaaliversiosi)",
"chooseInstrumentalPrompt": "Kuuntele ja valitse, mitä instrumentaaliversiota videossasi käytetään."
},
"searchLyrics": {
"artist": "Esittäjä",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Ta vidéo karaoké utilisera :",
"recommended": "recommandé",
"autoInstrumentalBacking": "Nous gardons les chœurs dans l'instrumental. Écoute bien : si tu entends la voix du chanteur principal, passe à « Instrumental propre » ci-dessus.",
"advancedMode": "Mode avancé (modifier les chœurs ou importer ton propre instrumental)"
"advancedMode": "Mode avancé (modifier les chœurs ou importer ton propre instrumental)",
"chooseInstrumentalPrompt": "Écoute et choisis l'instrumental à utiliser pour ta vidéo."
},
"searchLyrics": {
"artist": "Artiste",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "סרטון הקריוקי שלך יכלול:",
"recommended": "מומלץ",
"autoInstrumentalBacking": "אנחנו משאירים את קולות הרקע בגרסה האינסטרומנטלית. האזינו — אם אתם שומעים את קולו של הזמר הראשי, עברו ל״אינסטרומנטלי נקי״ למעלה.",
"advancedMode": "מצב מתקדם (עריכת קולות רקע או העלאת פלייבק משלך)"
"advancedMode": "מצב מתקדם (עריכת קולות רקע או העלאת פלייבק משלך)",
"chooseInstrumentalPrompt": "האזינו ובחרו באיזה קטע אינסטרומנטלי להשתמש בסרטון שלכם."
},
"searchLyrics": {
"artist": "אמן",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "तुम्हारे कराओके वीडियो में इस्तेमाल होगा:",
"recommended": "सुझाया गया",
"autoInstrumentalBacking": "हम इंस्ट्रुमेंटल में बैकिंग वोकल्स रख रहे हैं। इसे सुनकर देखो — अगर तुम्हें मुख्य गायक की आवाज़ सुनाई दे रही है, तो ऊपर “Clean instrumental” पर स्विच करो।",
"advancedMode": "एडवांस्ड मोड (बैकिंग वोकल्स एडिट करो या अपना खुद का इंस्ट्रूमेंटल अपलोड करो)"
"advancedMode": "एडवांस्ड मोड (बैकिंग वोकल्स एडिट करो या अपना खुद का इंस्ट्रूमेंटल अपलोड करो)",
"chooseInstrumentalPrompt": "सुनकर चुनो कि तुम्हारे वीडियो में कौन सा इंस्ट्रूमेंटल इस्तेमाल होना चाहिए।"
},
"searchLyrics": {
"artist": "आर्टिस्ट",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Tvoj karaoke video koristit će:",
"recommended": "preporučeno",
"autoInstrumentalBacking": "Zadržavamo prateće vokale u instrumentalu. Poslušaj — ako čuješ glas glavnog pjevača, prebaci na „Čisti instrumental” iznad.",
"advancedMode": "Napredni način rada (uredi prateće vokale ili prenesi vlastiti instrumental)"
"advancedMode": "Napredni način rada (uredi prateće vokale ili prenesi vlastiti instrumental)",
"chooseInstrumentalPrompt": "Poslušaj i odaberi koji instrumental želiš koristiti u svom videu."
},
"searchLyrics": {
"artist": "Izvođač",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "A karaoke videódhoz ezt fogjuk használni:",
"recommended": "ajánlott",
"autoInstrumentalBacking": "Megtartjuk a háttérvokálokat a zenei alapban. Hallgasd meg — ha behallatszik a főénekes hangja, válts a fenti „Tiszta zenei alap” lehetőségre.",
"advancedMode": "Haladó mód (szerkeszd a háttérvokálokat, vagy töltsd fel a saját zenei alapodat)"
"advancedMode": "Haladó mód (szerkeszd a háttérvokálokat, vagy töltsd fel a saját zenei alapodat)",
"chooseInstrumentalPrompt": "Hallgasd meg, és válaszd ki, hogy melyik zenei alapot használja a videód."
},
"searchLyrics": {
"artist": "Előadó",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Video karaoke Anda akan menggunakan:",
"recommended": "direkomendasikan",
"autoInstrumentalBacking": "Kami mempertahankan vokal latar pada instrumental. Coba dengarkan — jika Anda masih bisa mendengar suara penyanyi utama, silakan beralih ke “Instrumental bersih” di atas.",
"advancedMode": "Mode lanjutan (edit vokal latar atau unggah instrumental Anda sendiri)"
"advancedMode": "Mode lanjutan (edit vokal latar atau unggah instrumental Anda sendiri)",
"chooseInstrumentalPrompt": "Coba dengarkan dan pilih instrumental mana yang akan digunakan di video Anda."
},
"searchLyrics": {
"artist": "Artis",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Il tuo video karaoke utilizzerà:",
"recommended": "consigliato",
"autoInstrumentalBacking": "Stiamo mantenendo i cori nella base strumentale. Ascolta — se riesci a sentire la voce del cantante principale, passa a “Strumentale pulita” qui sopra.",
"advancedMode": "Modalità avanzata (modifica i cori o carica la tua base strumentale)"
"advancedMode": "Modalità avanzata (modifica i cori o carica la tua base strumentale)",
"chooseInstrumentalPrompt": "Ascolta e scegli quale base musicale utilizzare per il tuo video."
},
"searchLyrics": {
"artist": "Artista",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "カラオケ動画で使用する音源:",
"recommended": "おすすめ",
"autoInstrumentalBacking": "インストゥルメンタルにはバックボーカルを残しています。一度お聴きいただき、もしリードボーカルの声が聞こえる場合は、上の「クリーン・インストゥルメンタル」に切り替えてください。",
"advancedMode": "詳細モード(バックボーカルの編集、またはご自身の伴奏をアップロード)"
"advancedMode": "詳細モード(バックボーカルの編集、またはご自身の伴奏をアップロード)",
"chooseInstrumentalPrompt": "試聴して、動画に使用するインストゥルメンタルを選んでください。"
},
"searchLyrics": {
"artist": "アーティスト",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "노래방 비디오에 다음이 사용됩니다:",
"recommended": "추천",
"autoInstrumentalBacking": "반주에 백보컬을 남겨두었습니다. 한번 들어보시고 메인 보컬의 목소리가 들린다면, 상단의 “Clean instrumental”로 변경해 주세요.",
"advancedMode": "고급 모드 (백보컬 편집 또는 직접 반주 업로드)"
"advancedMode": "고급 모드 (백보컬 편집 또는 직접 반주 업로드)",
"chooseInstrumentalPrompt": "들어보시고 영상에 사용할 반주를 선택해 주세요."
},
"searchLyrics": {
"artist": "아티스트",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Video karaoke anda akan menggunakan:",
"recommended": "disyorkan",
"autoInstrumentalBacking": "Kami mengekalkan vokal latar dalam instrumental. Cuba dengar — jika suara penyanyi utama masih kedengaran, tukar kepada “Instrumental bersih” di atas.",
"advancedMode": "Mod lanjutan (sunting vokal latar atau muat naik instrumental anda sendiri)"
"advancedMode": "Mod lanjutan (sunting vokal latar atau muat naik instrumental anda sendiri)",
"chooseInstrumentalPrompt": "Dengarkan dan pilih instrumental mana yang patut digunakan untuk video anda."
},
"searchLyrics": {
"artist": "Artis",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Karaokevideoen din vil bruke:",
"recommended": "anbefalt",
"autoInstrumentalBacking": "Vi beholder bakgrunnsvokalen i instrumentalsporet. Hør på det — hvis du kan høre hovedvokalistens stemme, bytt til «Ren instrumental» ovenfor.",
"advancedMode": "Avansert modus (rediger bakgrunnsvokal eller last opp din egen instrumental)"
"advancedMode": "Avansert modus (rediger bakgrunnsvokal eller last opp din egen instrumental)",
"chooseInstrumentalPrompt": "Hør gjennom og velg hvilken instrumental videoen din skal bruke."
},
"searchLyrics": {
"artist": "Artist",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Je karaokevideo gebruikt:",
"recommended": "aanbevolen",
"autoInstrumentalBacking": "We laten de achtergrondzang in de instrumentale versie staan. Luister even — als je de stem van de leadzanger er nog doorheen hoort, kies dan hierboven voor “Schone instrumentale versie”.",
"advancedMode": "Geavanceerde modus (bewerk achtergrondzang of upload je eigen instrumentale versie)"
"advancedMode": "Geavanceerde modus (bewerk achtergrondzang of upload je eigen instrumentale versie)",
"chooseInstrumentalPrompt": "Luister even en kies welke instrumentale versie je voor je video wilt gebruiken."
},
"searchLyrics": {
"artist": "Artiest",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "W Twoim wideo karaoke wykorzystamy:",
"recommended": "zalecane",
"autoInstrumentalBacking": "Zostawiamy chórki w podkładzie. Posłuchaj — jeśli słyszysz przebijający się głos głównego wokalisty, przełącz na „Czysty podkład” powyżej.",
"advancedMode": "Tryb zaawansowany (edytuj chórki lub prześlij własny podkład)"
"advancedMode": "Tryb zaawansowany (edytuj chórki lub prześlij własny podkład)",
"chooseInstrumentalPrompt": "Posłuchaj i wybierz, którego podkładu chcesz użyć w swoim wideo."
},
"searchLyrics": {
"artist": "Wykonawca",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Seu vídeo de karaokê usará:",
"recommended": "recomendado",
"autoInstrumentalBacking": "Estamos mantendo os vocais de apoio no instrumental. Dê uma ouvida — se você conseguir ouvir a voz do vocalista principal, mude para “Instrumental limpo” acima.",
"advancedMode": "Modo avançado (edite os vocais de apoio ou envie seu próprio instrumental)"
"advancedMode": "Modo avançado (edite os vocais de apoio ou envie seu próprio instrumental)",
"chooseInstrumentalPrompt": "Ouça e escolha qual instrumental o seu vídeo deve usar."
},
"searchLyrics": {
"artist": "Artista",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Videoclipul tău de karaoke va folosi:",
"recommended": "recomandat",
"autoInstrumentalBacking": "Păstrăm vocile de fundal în instrumental. Ascultă — dacă se aude vocea solistului principal, schimbă pe „Instrumental curat” mai sus.",
"advancedMode": "Mod avansat (editează vocile de fundal sau încarcă propriul tău instrumental)"
"advancedMode": "Mod avansat (editează vocile de fundal sau încarcă propriul tău instrumental)",
"chooseInstrumentalPrompt": "Ascultă și alege ce instrumental să folosească videoclipul tău."
},
"searchLyrics": {
"artist": "Artist",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "В твоём караоке-видео будет использоваться:",
"recommended": "рекомендуется",
"autoInstrumentalBacking": "Мы оставляем бэк-вокал в инструментале. Послушай — если пробивается голос солиста, переключись на «Чистый инструментал» выше.",
"advancedMode": "Расширенный режим (редактируй бэк-вокал или загрузи свою минусовку)"
"advancedMode": "Расширенный режим (редактируй бэк-вокал или загрузи свою минусовку)",
"chooseInstrumentalPrompt": "Послушай и выбери, какой инструментал использовать для твоего видео."
},
"searchLyrics": {
"artist": "Исполнитель",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "V tvojom karaoke videu sa použije:",
"recommended": "odporúčané",
"autoInstrumentalBacking": "V inštrumentálnej verzii nechávame sprievodné vokály. Vypočuj si to — ak počuješ prenikať hlas hlavného speváka, prepni vyššie na „Čistý inštrumentál“.",
"advancedMode": "Pokročilý režim (uprav sprievodné vokály alebo nahraj vlastný inštrumentál)"
"advancedMode": "Pokročilý režim (uprav sprievodné vokály alebo nahraj vlastný inštrumentál)",
"chooseInstrumentalPrompt": "Vypočuj si to a vyber si, ktorý inštrumentál chceš vo svojom videu použiť."
},
"searchLyrics": {
"artist": "Interpret",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Din karaokevideo kommer att använda:",
"recommended": "rekommenderas",
"autoInstrumentalBacking": "Vi behåller bakgrundssången i instrumentalversionen. Lyssna — om du hör huvudsångarens röst tränga igenom, byt till ”Ren instrumental” ovan.",
"advancedMode": "Avancerat läge (redigera bakgrundssång eller ladda upp ditt eget instrumentalspår)"
"advancedMode": "Avancerat läge (redigera bakgrundssång eller ladda upp ditt eget instrumentalspår)",
"chooseInstrumentalPrompt": "Lyssna och välj vilken instrumentalversion din video ska använda."
},
"searchLyrics": {
"artist": "Artist",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "วิดีโอคาราโอเกะของคุณจะใช้:",
"recommended": "แนะนำ",
"autoInstrumentalBacking": "เราจะเก็บเสียงร้องประสานไว้ในดนตรีบรรเลง ลองฟังดูนะ — ถ้าคุณยังได้ยินเสียงนักร้องนำแทรกเข้ามา ให้เปลี่ยนไปเลือก “ดนตรีบรรเลงล้วน” ที่ด้านบน",
"advancedMode": "โหมดขั้นสูง (แก้ไขเสียงร้องประสานหรืออัปโหลดดนตรีบรรเลงของคุณเอง)"
"advancedMode": "โหมดขั้นสูง (แก้ไขเสียงร้องประสานหรืออัปโหลดดนตรีบรรเลงของคุณเอง)",
"chooseInstrumentalPrompt": "ลองฟังและเลือกดนตรีบรรเลงที่คุณต้องการใช้ในวิดีโอของคุณ"
},
"searchLyrics": {
"artist": "ศิลปิน",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Gagamitin sa iyong karaoke video ang:",
"recommended": "inirerekomenda",
"autoInstrumentalBacking": "Pinanatili namin ang mga backing vocal sa instrumental. Pakinggan mo — kung naririnig mo ang boses ng lead singer, lumipat sa “Clean instrumental” sa itaas.",
"advancedMode": "Advanced mode (i-edit ang backing vocals o i-upload ang sarili mong instrumental)"
"advancedMode": "Advanced mode (i-edit ang backing vocals o i-upload ang sarili mong instrumental)",
"chooseInstrumentalPrompt": "Pakinggan at piliin kung aling instrumental ang dapat gamitin sa iyong video."
},
"searchLyrics": {
"artist": "Artist",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Karaoke videonda şunlar kullanılacak:",
"recommended": "önerilen",
"autoInstrumentalBacking": "Enstrümantalde geri vokalleri koruyoruz. Bir dinle — eğer ana solistin sesini hala duyabiliyorsan, yukarıdaki “Temiz enstrümantal” seçeneğine geç.",
"advancedMode": "Gelişmiş mod (geri vokalleri düzenle veya kendi enstrümantalini yükle)"
"advancedMode": "Gelişmiş mod (geri vokalleri düzenle veya kendi enstrümantalini yükle)",
"chooseInstrumentalPrompt": "Dinle ve videon için hangi enstrümantali kullanacağını seç."
},
"searchLyrics": {
"artist": "Sanatçı",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "У твоєму караоке-відео буде використано:",
"recommended": "рекомендовано",
"autoInstrumentalBacking": "Ми залишаємо бек-вокал в інструменталі. Послухай — якщо чуєш голос соліста, перемкнись на «Чистий інструментал» вище.",
"advancedMode": "Розширений режим (редагуй бек-вокал або завантаж власну інструментальну версію)"
"advancedMode": "Розширений режим (редагуй бек-вокал або завантаж власну інструментальну версію)",
"chooseInstrumentalPrompt": "Послухай та обери, яку інструментальну версію слід використати для твого відео."
},
"searchLyrics": {
"artist": "Виконавець",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "Video karaoke của bạn sẽ sử dụng:",
"recommended": "khuyên dùng",
"autoInstrumentalBacking": "Chúng tôi đang giữ lại phần hát bè trong bản nhạc nền. Hãy nghe thử nhé — nếu bạn vẫn nghe thấy giọng ca sĩ chính, hãy chuyển sang “Nhạc nền sạch” ở bên trên.",
"advancedMode": "Chế độ nâng cao (chỉnh sửa giọng bè hoặc tải lên nhạc nền của riêng bạn)"
"advancedMode": "Chế độ nâng cao (chỉnh sửa giọng bè hoặc tải lên nhạc nền của riêng bạn)",
"chooseInstrumentalPrompt": "Hãy nghe thử và chọn bản nhạc nền cho video của bạn."
},
"searchLyrics": {
"artist": "Nghệ sĩ",
+2 -1
View File
@@ -1161,7 +1161,8 @@
"finalChoiceLabel": "您的卡拉OK视频将使用:",
"recommended": "推荐",
"autoInstrumentalBacking": "我们在伴奏中保留了伴唱。请试听一下——如果您能听到主唱的声音,请切换到上方的“纯净伴奏”。",
"advancedMode": "高级模式(编辑伴唱或上传您自己的伴奏)"
"advancedMode": "高级模式(编辑伴唱或上传您自己的伴奏)",
"chooseInstrumentalPrompt": "试听一下,然后选择您的视频要使用的伴奏。"
},
"searchLyrics": {
"artist": "歌手",
+1 -1
View File
@@ -1,6 +1,6 @@
[tool.poetry]
name = "karaoke-gen"
version = "0.213.1"
version = "0.214.0"
description = "Generate karaoke videos with synchronized lyrics. Handles the entire process from downloading audio and lyrics to creating the final video with title screens."
authors = ["Andrew Beveridge <andrew@beveridge.uk>"]
license = "MIT"