Skip to content

df/io: Add support for torchaudio 2.9. #670

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
lubosz wants to merge 1 commit into
base: main
Choose a base branch
from
Open

df/io: Add support for torchaudio 2.9. #670

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
45 changes: 34 additions & 11 deletions DeepFilterNet/df/io.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,25 +6,38 @@
from loguru import logger
from numpy import ndarray
from torch import Tensor
from packaging import version

try:
from torchaudio import AudioMetaData

if version.parse(ta.__version__) >= version.parse("2.9.0"):
from torchcodec.decoders import AudioDecoder
from torchcodec._core import AudioStreamMetadata

AudioStreamMetadataType = AudioStreamMetadata

TA_RESAMPLE_SINC = "sinc_interp_hann"
TA_RESAMPLE_KAISER = "sinc_interp_kaiser"
except ImportError:
from torchaudio.backend.common import AudioMetaData
else:
try:
from torchaudio import AudioMetaData

TA_RESAMPLE_SINC = "sinc_interp_hann"
TA_RESAMPLE_KAISER = "sinc_interp_kaiser"
except ImportError:
from torchaudio.backend.common import AudioMetaData

TA_RESAMPLE_SINC = "sinc_interpolation"
TA_RESAMPLE_KAISER = "kaiser_window"
TA_RESAMPLE_SINC = "sinc_interpolation"
TA_RESAMPLE_KAISER = "kaiser_window"

AudioStreamMetadataType = AudioMetaData

from df.logger import warn_once
from df.utils import download_file, get_cache_dir, get_git_root


def load_audio(
file: str, sr: Optional[int] = None, verbose=True, **kwargs
) -> Tuple[Tensor, AudioMetaData]:
) -> Tuple[Tensor, AudioStreamMetadataType]:
"""Loads an audio file using torchaudio.

Args:
Expand All @@ -43,10 +56,20 @@ def load_audio(
rkwargs = {}
if "method" in kwargs:
rkwargs["method"] = kwargs.pop("method")
info: AudioMetaData = ta.info(file, **ikwargs)
if "num_frames" in kwargs and sr is not None:
kwargs["num_frames"] *= info.sample_rate // sr
audio, orig_sr = ta.load(file, **kwargs)

if version.parse(ta.__version__) >= version.parse("2.9.0"):
decoder = AudioDecoder(file)
info: AudioStreamMetadata = decoder.metadata
samples = decoder.get_all_samples()
audio = samples.data
orig_sr = samples.sample_rate
else:
info: AudioMetaData = ta.info(file, **ikwargs)

if "num_frames" in kwargs and sr is not None:
kwargs["num_frames"] *= info.sample_rate // sr
audio, orig_sr = ta.load(file, **kwargs)

if sr is not None and orig_sr != sr:
if verbose:
warn_once(
Expand Down