Update main.py
Browse files
main.py
CHANGED
|
@@ -5,29 +5,41 @@ import os
|
|
| 5 |
app = Flask(__name__)
|
| 6 |
|
| 7 |
# Load the Whisper model
|
| 8 |
-
|
|
|
|
|
|
|
| 9 |
|
| 10 |
def transcribe(audio_path):
|
|
|
|
|
|
|
| 11 |
# Load audio and pad/trim it to fit 30 seconds
|
|
|
|
| 12 |
audio = whisper.load_audio(audio_path)
|
| 13 |
audio = whisper.pad_or_trim(audio)
|
| 14 |
|
| 15 |
# Make log-Mel spectrogram and move to the same device as the model
|
|
|
|
| 16 |
mel = whisper.log_mel_spectrogram(audio).to(model.device)
|
| 17 |
|
| 18 |
# Detect the spoken language
|
|
|
|
| 19 |
_, probs = model.detect_language(mel)
|
| 20 |
language = max(probs, key=probs.get)
|
| 21 |
-
|
|
|
|
| 22 |
# Decode the audio
|
|
|
|
| 23 |
options = whisper.DecodingOptions(fp16=False)
|
| 24 |
result = whisper.decode(model, mel, options)
|
| 25 |
|
|
|
|
| 26 |
return result.text, language
|
| 27 |
|
| 28 |
@app.route('/transcribe', methods=['POST'])
|
| 29 |
def transcribe_audio():
|
|
|
|
| 30 |
if 'audio' not in request.files:
|
|
|
|
| 31 |
return jsonify({"error": "No audio file provided"}), 400
|
| 32 |
|
| 33 |
audio_file = request.files['audio']
|
|
@@ -36,22 +48,28 @@ def transcribe_audio():
|
|
| 36 |
audio_path = os.path.join("temp_audio", audio_file.filename)
|
| 37 |
os.makedirs("temp_audio", exist_ok=True)
|
| 38 |
audio_file.save(audio_path)
|
|
|
|
| 39 |
|
| 40 |
# Transcribe the audio
|
| 41 |
try:
|
| 42 |
transcription, language = transcribe(audio_path)
|
| 43 |
except Exception as e:
|
|
|
|
| 44 |
return jsonify({"error": f"An error occurred: {str(e)}"}), 500
|
| 45 |
|
| 46 |
# Clean up the saved file
|
| 47 |
os.remove(audio_path)
|
|
|
|
| 48 |
|
| 49 |
# Return the transcription and detected language
|
|
|
|
| 50 |
return jsonify({"transcription": transcription, "language": language}), 200
|
| 51 |
|
| 52 |
@app.route('/healthcheck', methods=['GET'])
|
| 53 |
def healthcheck():
|
|
|
|
| 54 |
return jsonify({"status": "API is running"}), 200
|
| 55 |
|
| 56 |
if __name__ == '__main__':
|
|
|
|
| 57 |
app.run(host="0.0.0.0", port=5000)
|
|
|
|
| 5 |
app = Flask(__name__)
|
| 6 |
|
| 7 |
# Load the Whisper model
|
| 8 |
+
print("Loading Whisper model...")
|
| 9 |
+
model = whisper.load_model("tiny")
|
| 10 |
+
print("Whisper model loaded.")
|
| 11 |
|
| 12 |
def transcribe(audio_path):
|
| 13 |
+
print(f"Transcribing audio from: {audio_path}")
|
| 14 |
+
|
| 15 |
# Load audio and pad/trim it to fit 30 seconds
|
| 16 |
+
print("Loading and processing audio...")
|
| 17 |
audio = whisper.load_audio(audio_path)
|
| 18 |
audio = whisper.pad_or_trim(audio)
|
| 19 |
|
| 20 |
# Make log-Mel spectrogram and move to the same device as the model
|
| 21 |
+
print("Creating log-Mel spectrogram...")
|
| 22 |
mel = whisper.log_mel_spectrogram(audio).to(model.device)
|
| 23 |
|
| 24 |
# Detect the spoken language
|
| 25 |
+
print("Detecting language...")
|
| 26 |
_, probs = model.detect_language(mel)
|
| 27 |
language = max(probs, key=probs.get)
|
| 28 |
+
print(f"Detected language: {language}")
|
| 29 |
+
|
| 30 |
# Decode the audio
|
| 31 |
+
print("Decoding audio...")
|
| 32 |
options = whisper.DecodingOptions(fp16=False)
|
| 33 |
result = whisper.decode(model, mel, options)
|
| 34 |
|
| 35 |
+
print("Transcription complete.")
|
| 36 |
return result.text, language
|
| 37 |
|
| 38 |
@app.route('/transcribe', methods=['POST'])
|
| 39 |
def transcribe_audio():
|
| 40 |
+
print("Received request at /transcribe")
|
| 41 |
if 'audio' not in request.files:
|
| 42 |
+
print("Error: No audio file provided")
|
| 43 |
return jsonify({"error": "No audio file provided"}), 400
|
| 44 |
|
| 45 |
audio_file = request.files['audio']
|
|
|
|
| 48 |
audio_path = os.path.join("temp_audio", audio_file.filename)
|
| 49 |
os.makedirs("temp_audio", exist_ok=True)
|
| 50 |
audio_file.save(audio_path)
|
| 51 |
+
print(f"Audio file saved to: {audio_path}")
|
| 52 |
|
| 53 |
# Transcribe the audio
|
| 54 |
try:
|
| 55 |
transcription, language = transcribe(audio_path)
|
| 56 |
except Exception as e:
|
| 57 |
+
print(f"Error during transcription: {str(e)}")
|
| 58 |
return jsonify({"error": f"An error occurred: {str(e)}"}), 500
|
| 59 |
|
| 60 |
# Clean up the saved file
|
| 61 |
os.remove(audio_path)
|
| 62 |
+
print(f"Audio file removed from: {audio_path}")
|
| 63 |
|
| 64 |
# Return the transcription and detected language
|
| 65 |
+
print(f"Transcription: {transcription}, Language: {language}")
|
| 66 |
return jsonify({"transcription": transcription, "language": language}), 200
|
| 67 |
|
| 68 |
@app.route('/healthcheck', methods=['GET'])
|
| 69 |
def healthcheck():
|
| 70 |
+
print("Received request at /healthcheck")
|
| 71 |
return jsonify({"status": "API is running"}), 200
|
| 72 |
|
| 73 |
if __name__ == '__main__':
|
| 74 |
+
print("Starting Flask app...")
|
| 75 |
app.run(host="0.0.0.0", port=5000)
|