Spaces:
Sleeping
Sleeping
Removed processFile()
Browse files
utils.py
CHANGED
|
@@ -34,55 +34,6 @@ TRANSPARENT_BG = dict(
|
|
| 34 |
paper_bgcolor="rgba(0,0,0,0)",
|
| 35 |
)
|
| 36 |
|
| 37 |
-
# ---------------------------------------------------------------------------
|
| 38 |
-
# Audio processing
|
| 39 |
-
# ---------------------------------------------------------------------------
|
| 40 |
-
|
| 41 |
-
def processFile(filePath, pipeline, enableDenoise, earlyCleanup,
|
| 42 |
-
gainWindow, minimumGain, maximumGain,
|
| 43 |
-
dfModel=None, dfState=None, attenLimDB=3):
|
| 44 |
-
"""Load, optionally denoise, equalize, and diarize an audio file.
|
| 45 |
-
|
| 46 |
-
Returns (annotations, totalTimeInSeconds, waveform_tensor, sample_rate).
|
| 47 |
-
"""
|
| 48 |
-
print("Loading file")
|
| 49 |
-
waveformList, sampleRate = su.splitIntoTimeSegments(filePath, 600)
|
| 50 |
-
print("File loaded")
|
| 51 |
-
|
| 52 |
-
enhancedWaveformList = []
|
| 53 |
-
if enableDenoise:
|
| 54 |
-
print("Denoising")
|
| 55 |
-
for w in waveformList:
|
| 56 |
-
if enableDenoise:
|
| 57 |
-
from df import enhance
|
| 58 |
-
newW = enhance(dfModel, dfState, w, atten_lim_db=attenLimDB).detach().cpu()
|
| 59 |
-
enhancedWaveformList.append(newW)
|
| 60 |
-
else:
|
| 61 |
-
enhancedWaveformList.append(w)
|
| 62 |
-
if enableDenoise:
|
| 63 |
-
print("Audio denoised")
|
| 64 |
-
|
| 65 |
-
waveformEnhanced = su.combineWaveforms(enhancedWaveformList)
|
| 66 |
-
if earlyCleanup:
|
| 67 |
-
del enhancedWaveformList
|
| 68 |
-
|
| 69 |
-
print("Equalizing Audio")
|
| 70 |
-
waveform_gain_adjusted = su.equalizeVolume()(
|
| 71 |
-
waveformEnhanced, sampleRate, gainWindow, minimumGain, maximumGain
|
| 72 |
-
)
|
| 73 |
-
if earlyCleanup:
|
| 74 |
-
del waveformEnhanced
|
| 75 |
-
print("Audio Equalized")
|
| 76 |
-
|
| 77 |
-
print("Detecting speakers")
|
| 78 |
-
diarization_output = pipeline({"waveform": waveform_gain_adjusted, "sample_rate": sampleRate})
|
| 79 |
-
annotations = diarization_output.speaker_diarization
|
| 80 |
-
print("Speakers Detected")
|
| 81 |
-
|
| 82 |
-
totalTimeInSeconds = int(waveform_gain_adjusted.shape[-1] / sampleRate)
|
| 83 |
-
return annotations, totalTimeInSeconds, waveform_gain_adjusted, sampleRate
|
| 84 |
-
|
| 85 |
-
|
| 86 |
# ---------------------------------------------------------------------------
|
| 87 |
# Speaker sample helpers
|
| 88 |
# ---------------------------------------------------------------------------
|
|
|
|
| 34 |
paper_bgcolor="rgba(0,0,0,0)",
|
| 35 |
)
|
| 36 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
# ---------------------------------------------------------------------------
|
| 38 |
# Speaker sample helpers
|
| 39 |
# ---------------------------------------------------------------------------
|