Hackathons/HooHacks-12
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Demo Fixes 4

Browse Source
This commit is contained in:
GamerBoss101
2025-03-30 02:31:58 -04:00
parent fbb3ff4006
commit 6622d0c605
1 changed files with 7 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
Backend/server.py
View File

@@ -71,13 +71,15 @@ def load_models():
logger.info("Loading ASR pipeline...") logger.info("Loading ASR pipeline...")
try: try:
# Initialize the pipeline without the language parameter in the constructor
models.asr = pipeline( models.asr = pipeline(
"automatic-speech-recognition", "automatic-speech-recognition",
model="openai/whisper-small", model="openai/whisper-small",
device=DEVICE, device=DEVICE
language="en", # Force English language
return_attention_mask=True # Add attention mask
) )
# Configure the model with the appropriate options
# Note that for whisper, language should be set during inference, not initialization
logger.info("ASR pipeline loaded successfully") logger.info("ASR pipeline loaded successfully")
socketio.emit('model_status', {'model': 'asr', 'status': 'loaded'}) socketio.emit('model_status', {'model': 'asr', 'status': 'loaded'})
except Exception as e: except Exception as e:
@@ -312,7 +314,8 @@ def process_audio_and_respond(session_id, data):
# Use the ASR pipeline to transcribe # Use the ASR pipeline to transcribe
transcription_result = models.asr( transcription_result = models.asr(
{"array": waveform.squeeze().cpu().numpy(), "sampling_rate": models.generator.sample_rate}, {"array": waveform.squeeze().cpu().numpy(), "sampling_rate": models.generator.sample_rate},
return_timestamps=False return_timestamps=False,
generate_kwargs={"language": "en"} # Set language during inference
) )
user_text = transcription_result['text'].strip() user_text = transcription_result['text'].strip()