def prepare_audio_single(audio_paths: List[Pathlike], ) -> RecordingSet:
import soundfile as sf
recordings = []
for audio_path in tqdm(audio_paths, desc="Preparing audio"):
session_name = audio_path.parts[-2]
if audio_path.suffix == ".wav":
audio_sf = sf.SoundFile(str(audio_path))
num_frames = audio_sf.frames
num_channels = audio_sf.channels
samplerate = audio_sf.samplerate
else:
audio_sf, samplerate = read_sph(audio_path)
num_channels, num_frames = audio_sf.shape
recordings.append(
Recording(
id=session_name,
sources=[
AudioSource(
type="file",
channels=list(range(num_channels)),
source=str(audio_path),
)
],
sampling_rate=samplerate,
num_samples=num_frames,
duration=num_frames / samplerate,
))
return RecordingSet.from_recordings(recordings)
def prepare_audio_grouped(
audio_paths: List[Pathlike],
channel_to_idx_map: Dict[str, Dict[str, int]] = None,
) -> RecordingSet:
# Group together multiple channels from the same session.
# We will use that to create a Recording with multiple sources (channels).
from cytoolz import groupby
channel_wavs = groupby(lambda p: p.parts[-2], audio_paths)
if channel_to_idx_map is None:
channel_to_idx_map = defaultdict(dict)
recordings = []
for session_name, channel_paths in tqdm(channel_wavs.items(),
desc="Preparing audio"):
if session_name not in channel_to_idx_map:
channel_to_idx_map[session_name] = {
c: idx
for idx, c in enumerate(["chanE", "chanF", "chan6", "chan7"])
}
audio_sf, samplerate = read_sph(channel_paths[0])
recordings.append(
Recording(
id=session_name,
sources=[
AudioSource(
type="file",
channels=[
channel_to_idx_map[session_name][audio_path.stem]
],
source=str(audio_path),
) for audio_path in sorted(channel_paths)
if audio_path.stem in channel_to_idx_map[session_name]
],
sampling_rate=samplerate,
num_samples=audio_sf.shape[1],
duration=audio_sf.shape[1] / samplerate,
))
return RecordingSet.from_recordings(recordings)