admin
109 lines
3.0 KiB
Python
109 lines
3.0 KiB
Python
"""
|
|
Audio Processor Module
|
|
=======================
|
|
|
|
Simplified audio processor for ScrAIbe.
|
|
|
|
Previously this used torch and pyannote-style processing. In the LocalAI-backed
|
|
version, we primarily pass files to the API, but we keep a lightweight helper
|
|
for backward compatibility.
|
|
"""
|
|
|
|
from subprocess import CalledProcessError, run
|
|
import numpy as np
|
|
|
|
SAMPLE_RATE = 16000
|
|
NORMALIZATION_FACTOR = 32768.0
|
|
|
|
|
|
class AudioProcessor:
|
|
"""
|
|
Lightweight audio processor for loading and cutting audio.
|
|
|
|
Attributes:
|
|
waveform (np.ndarray): The audio waveform as float32.
|
|
sr (int): The sample rate of the audio.
|
|
"""
|
|
|
|
def __init__(self, waveform: np.ndarray, sr: int = SAMPLE_RATE):
|
|
self.waveform = waveform
|
|
self.sr = sr
|
|
|
|
if not isinstance(self.sr, int):
|
|
raise ValueError(
|
|
"Sample rate should be a single value of type int, "
|
|
f"not {len(self.sr)} and type {type(self.sr)}"
|
|
)
|
|
|
|
@classmethod
|
|
def from_file(cls, file: str, *args, **kwargs):
|
|
"""
|
|
Create an AudioProcessor instance from an audio file.
|
|
|
|
Args:
|
|
file (str): The audio file path.
|
|
|
|
Returns:
|
|
AudioProcessor: Instance with loaded audio.
|
|
"""
|
|
audio, sr = cls.load_audio(file, *args, **kwargs)
|
|
return cls(audio, sr)
|
|
|
|
def cut(self, start: float, end: float) -> np.ndarray:
|
|
"""
|
|
Cut a segment from the audio waveform.
|
|
|
|
Args:
|
|
start (float): Start time in seconds.
|
|
end (float): End time in seconds.
|
|
|
|
Returns:
|
|
np.ndarray: The cut waveform segment.
|
|
"""
|
|
start_idx = int(start * self.sr)
|
|
end_idx = int(np.ceil(end * self.sr))
|
|
return self.waveform[start_idx:end_idx]
|
|
|
|
@staticmethod
|
|
def load_audio(file: str, sr: int = SAMPLE_RATE):
|
|
"""
|
|
Load an audio file as a mono waveform, resampling if necessary.
|
|
Requires ffmpeg in PATH.
|
|
|
|
Args:
|
|
file (str): The audio file to open.
|
|
sr (int, optional): The desired sample rate.
|
|
|
|
Returns:
|
|
tuple: (waveform as np.ndarray[float32], sample rate)
|
|
|
|
Raises:
|
|
RuntimeError: If failed to load audio.
|
|
"""
|
|
cmd = [
|
|
"ffmpeg",
|
|
"-nostdin",
|
|
"-threads", "0",
|
|
"-i", file,
|
|
"-f", "s16le",
|
|
"-ac", "1",
|
|
"-acodec", "pcm_s16le",
|
|
"-ar", str(sr),
|
|
"-",
|
|
]
|
|
try:
|
|
out = run(cmd, capture_output=True, check=True).stdout
|
|
except CalledProcessError as e:
|
|
raise RuntimeError(
|
|
f"Failed to load audio: {e.stderr.decode()}"
|
|
) from e
|
|
|
|
waveform = np.frombuffer(out, np.int16).flatten().astype(
|
|
np.float32
|
|
) / NORMALIZATION_FACTOR
|
|
|
|
return waveform, sr
|
|
|
|
def __repr__(self) -> str:
|
|
return f"AudioProcessor(waveform_len={len(self.waveform)}, sr={self.sr})"
|