Files
pyannote-audio/tests/io_test.py
T
ea18bc70d8 Align audio CI and releases; require Python 3.11 (#2081)
* feat(cli): add --subfolder option to benchmark command (#2063)
* fix(utils): fix `SpeakerDiarizationMixin.optimal_mapping` when `reference` is a mapping
* fix(task): fix `balance` option crashing training with `KeyError: 'metadata'` (#2067)
* fix(io): fix `Audio.crop` raising when an in-memory waveform is cropped up to its very end (#2068)
* fix(model): make Model.default_activation return identity for unsupported tasks
* fix: compute sample duration from preloaded waveform
* fix(pipeline): omit unset runtime options from model references (#2080)

Co-authored-by: abeeranajam31 <128603901+abeeranajam31@users.noreply.github.com>
Co-authored-by: antoinelaurent <antoinelaurent@me.com>
Co-authored-by: Kayvan Zahiri <123409429+Kayvan-Zahiri@users.noreply.github.com>
Co-authored-by: Mohammad Hijjawi <51020382+MohammadHijjawi97@users.noreply.github.com>
2026-10-22 15:07:01 +02:00

96 lines
2.8 KiB
Python

import torch
from pyannote.core import Segment
from torch import Tensor
from torchcodec.decoders import AudioDecoder, AudioStreamMetadata
from pyannote.audio.core.io import Audio
def test_audio_resample():
"Audio is correctly resampled when it isn't the correct sample rate"
test_file = "tests/data/dev00.wav"
metadata: AudioStreamMetadata = AudioDecoder(test_file).metadata
old_sr = metadata.sample_rate
loader = Audio(sample_rate=old_sr // 2, mono="downmix")
wav, sr = loader(test_file)
assert isinstance(wav, Tensor)
assert sr == old_sr // 2
def test_basic_load_with_defaults():
test_file = "tests/data/dev00.wav"
loader = Audio(mono="downmix")
wav, _sr = loader(test_file)
assert isinstance(wav, Tensor)
def test_correct_audio_channel():
"When we specify an audio channel, it is chosen correctly"
waveform = torch.rand(2, 16000 * 2)
loader = Audio(mono="downmix")
wav, sr = loader({"waveform": waveform, "sample_rate": 16000, "channel": 1})
assert torch.equal(wav, waveform[1:2])
assert sr == 16000
def test_can_load_with_waveform():
"We can load a raw waveform"
waveform = torch.rand(2, 16000 * 2)
loader = Audio(mono="downmix")
wav, sr = loader({"waveform": waveform, "sample_rate": 16000})
assert isinstance(wav, Tensor)
assert sr == 16000
def test_can_crop():
"Cropping works when we give a Segment"
test_file = "tests/data/dev00.wav"
loader = Audio(mono="downmix")
segment = Segment(0.2, 0.7)
wav, sr = loader.crop(test_file, segment)
assert wav.shape[1] / sr == 0.5
def test_can_crop_waveform():
"Cropping works on raw waveforms"
waveform = torch.rand(1, 16000 * 2)
loader = Audio(mono="downmix")
segment = Segment(0.2, 0.7)
wav, sr = loader.crop({"waveform": waveform, "sample_rate": 16000}, segment)
assert isinstance(wav, Tensor)
assert sr == 16000
def test_can_crop_waveform_up_to_its_end():
"Cropping a raw waveform up to its very last sample does not raise"
waveform = torch.rand(1, 16000 * 2)
loader = Audio(mono="downmix")
segment = Segment(1.5, 2.0)
wav, _sr = loader.crop({"waveform": waveform, "sample_rate": 16000}, segment)
assert torch.equal(wav, waveform[:, 24000:])
# File Like Object Tests
def test_can_load_from_file_like():
"Load entire wav of file like"
loader = Audio(mono="downmix")
with open("tests/data/dev00.wav", "rb") as f:
wav, sr = loader(f)
assert isinstance(wav, Tensor)
assert sr == 16000
def test_can_crop_from_file_like():
"Load cropped sections from file like objects"
loader = Audio(mono="downmix")
with open("tests/data/dev00.wav", "rb") as f:
segment = Segment(0.2, 0.7)
wav, sr = loader.crop(f, segment)
assert isinstance(wav, Tensor)
assert sr == 16000
assert wav.shape[1] == 0.5 * 16000