Auto backup: 2026-02-21 07:01

.venvs/transcribe/transcribe_vosk.py

@@ -0,0 +1,22 @@
+import sys, json, wave, subprocess, os
+from vosk import Model, KaldiRecognizer
+
+audio_in=sys.argv[1]
+model_path='/home/openclaw/.openclaw/workspace/.cache/vosk/vosk-model-small-en-us-0.15'
+wav='/tmp/in.wav'
+subprocess.check_call(['ffmpeg','-y','-i',audio_in,'-ar','16000','-ac','1','-f','wav',wav], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+wf=wave.open(wav,'rb')
+model=Model(model_path)
+rec=KaldiRecognizer(model, wf.getframerate())
+rec.SetWords(True)
+text=[]
+while True:
+    data=wf.readframes(4000)
+    if len(data)==0:
+        break
+    if rec.AcceptWaveform(data):
+        r=json.loads(rec.Result())
+        if r.get('text'): text.append(r['text'])
+r=json.loads(rec.FinalResult())
+if r.get('text'): text.append(r['text'])
+print(' '.join(text).strip())