diar

app.py

@@ -0,0 +1,23 @@
+import os
+import gradio as gr
+from scipy.io.wavfile import write
+from simple_diarizer.diarizer import Diarizer
+from simple_diarizer.utils import (check_wav_16khz_mono, convert_wavfile)
+
+
+
+def inference(audio):
+    os.makedirs("out", exist_ok=True)
+    write('test.wav', audio[0], audio[1])
+    os.system("python3 -m demucs.separate -n htdemucs --two-stems=vocals -d cpu test.wav -o out")
+    return "./out/htdemucs/test/vocals.wav", "./out/htdemucs/test/no_vocals.wav"
+
+
+title = "audio_denoise and speakser diarization"
+
+gr.Interface(
+    inference,
+    gr.Audio(type="numpy", label="Input"),
+    [gr.Audio(type="filepath", label="Vocal"), gr.Audio(type="filepath", label="No Vocals / Instrumental")],
+    title=title,
+).launch(enable_queue=True)

packages.txt

@@ -0,0 +1 @@
+ffmpeg

requirements.txt

@@ -0,0 +1,21 @@
+git+https://github.com/openai/whisper.git
+demucs
+pydub
+gradio==3.12
+ffmpeg-python
+torch
+torchaudio
+tqdm==4.64.1
+EasyNMT==2.0.2   
+nltk
+transformers
+pysrt
+psutil==5.9.2
+requests
+gpuinfo
+faster-whisper
+yt-dlp
+lightning_fabric
+modelscope
+rotary_embedding_torch
+simple-diarizer

utils/diarization_pipeline.py

@@ -0,0 +1,34 @@
+from simple_diarizer.diarizer import Diarizer
+from simple_diarizer.utils import (check_wav_16khz_mono, convert_wavfile)
+
+import soundfile as sf
+
+
+
+
+class DiarizationPipeline:
+    def __init__(self, mode='torch'):
+        super(DiarizationPipeline, self).__init__()
+        self.diar = Diarizer(
+            embed_model='ecapa',  # supported types: ['xvec', 'ecapa']
+            cluster_method='ahc',  # supported types: ['ahc', 'sc']
+            window=1,  # size of window to extract embeddings (in seconds)
+            period=0.1  # hop of window (in seconds)
+        )
+        self.diar
+
+    def __call__(self, wav_file):
+        # wav_file = convert_wavfile(in_file, f"{outdir}/{YOUTUBE_ID}_converted.wav")
+
+        # signal, fs = sf.read(wav_file)
+        #
+        segments = self.diar.diarize(wav_file,
+                                     num_speakers=None,
+                                     threshold=9e-1,)
+
+        return segments
+
+
+if __name__ == '__main__':
+    pipeline = DiarizationPipeline('torch')
+    pipeline('path_audio')