Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -35,6 +35,7 @@ def process_audio(audio_bytes):
|
|
35 |
waveform = resampler(waveform)
|
36 |
return {"raw": waveform.numpy().squeeze(), "sampling_rate": 16000}
|
37 |
|
|
|
38 |
def get_voice_transcription(state_key):
|
39 |
"""Display audio recorder for a given key.
|
40 |
If new audio is recorded, transcribe it and update the session state.
|
@@ -51,17 +52,19 @@ def get_voice_transcription(state_key):
|
|
51 |
current_hash = hashlib.md5(audio_bytes).hexdigest()
|
52 |
last_hash_key = state_key + "_last_hash"
|
53 |
if st.session_state.get(last_hash_key, "") != current_hash:
|
54 |
-
st.
|
|
|
55 |
try:
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
|
|
65 |
return st.session_state[state_key]
|
66 |
|
67 |
######################################
|
|
|
35 |
waveform = resampler(waveform)
|
36 |
return {"raw": waveform.numpy().squeeze(), "sampling_rate": 16000}
|
37 |
|
38 |
+
|
39 |
def get_voice_transcription(state_key):
|
40 |
"""Display audio recorder for a given key.
|
41 |
If new audio is recorded, transcribe it and update the session state.
|
|
|
52 |
current_hash = hashlib.md5(audio_bytes).hexdigest()
|
53 |
last_hash_key = state_key + "_last_hash"
|
54 |
if st.session_state.get(last_hash_key, "") != current_hash:
|
55 |
+
st.session_sta
|
56 |
+
te[last_hash_key] = current_hash
|
57 |
try:
|
58 |
+
with st.spinner('π Processing your voice...'):
|
59 |
+
audio_input = process_audio(audio_bytes)
|
60 |
+
whisper = load_voice_model()
|
61 |
+
transcribed_text = whisper(audio_input)["text"]
|
62 |
+
st.info(f"π Transcribed: {transcribed_text}")
|
63 |
+
# Append (or set) new transcription
|
64 |
+
st.session_state[state_key] += (" " + transcribed_text).strip()
|
65 |
+
st.experimental_rerun()
|
66 |
+
except Exception as e:
|
67 |
+
st.error(f"Voice input error: {str(e)}")
|
68 |
return st.session_state[state_key]
|
69 |
|
70 |
######################################
|