MCPcopy Create free account
hub / github.com/FireRedTeam/FireRedTTS / load_audio

Function load_audio

fireredtts/utils/utils.py:9–37  ·  view source on GitHub ↗

_summary_ Args: audiopath (_type_): audio_path sampling_rate (_type_): sampling_rate Returns: _type_: _description_

(audiopath, sampling_rate)

Source from the content-addressed store, hash-verified

7
8
9def load_audio(audiopath, sampling_rate):
10 """_summary_
11
12 Args:
13 audiopath (_type_): audio_path
14 sampling_rate (_type_): sampling_rate
15
16 Returns:
17 _type_: _description_
18 """
19 audio, lsr = torchaudio.load(audiopath)
20
21 # stereo to mono if needed
22 if audio.size(0) != 1:
23 audio = torch.mean(audio, dim=0, keepdim=True)
24
25 # resample
26 audio_resampled = torchaudio.functional.resample(audio, lsr, sampling_rate)
27 if torch.any(audio > 10) or not torch.any(audio < 0):
28 print(f"Error with {audiopath}. Max={audio.max()} min={audio.min()}")
29
30 if torch.any(audio_resampled > 10) or not torch.any(audio_resampled < 0):
31 print(
32 f"Error with {audiopath}. Max={audio_resampled.max()} min={audio_resampled.min()}"
33 )
34 # clip audio invalid values
35 audio.clip_(-1, 1)
36 audio_resampled.clip_(-1, 1)
37 return audio, lsr, audio_resampled

Callers 1

Calls

no outgoing calls

Tested by

no test coverage detected