adelorenzo/talk2me
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

working version

Browse Source
This commit is contained in:
Adolfo Delorenzo 2025-04-07 09:18:05 -06:00
parent 1043402516
commit 8c8600d458
1 changed files with 36 additions and 36 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

72
app.py
View File

@ -28,7 +28,7 @@ def root_files(filename):
'apple-touch-icon-120x120.png', 'apple-touch-icon-120x120.png',
'apple-touch-icon-120x120-precomposed.png' 'apple-touch-icon-120x120-precomposed.png'
] ]
if filename in common_icons: if filename in common_icons:
# Map to appropriate icon in static/icons # Map to appropriate icon in static/icons
icon_mapping = { icon_mapping = {
@ -38,9 +38,9 @@ def root_files(filename):
'apple-touch-icon-120x120.png': 'apple-icon-120x120.png', 'apple-touch-icon-120x120.png': 'apple-icon-120x120.png',
'apple-touch-icon-120x120-precomposed.png': 'apple-icon-120x120.png' 'apple-touch-icon-120x120-precomposed.png': 'apple-icon-120x120.png'
} }
return send_from_directory('static/icons', icon_mapping.get(filename, 'apple-icon-180x180.png')) return send_from_directory('static/icons', icon_mapping.get(filename, 'apple-icon-180x180.png'))
# If not an icon, return 404 # If not an icon, return 404
return "File not found", 404 return "File not found", 404
@ -206,11 +206,11 @@ LANGUAGE_TO_CODE = {v: k for k, v in SUPPORTED_LANGUAGES.items()}
# Map language names to OpenAI TTS voice options # Map language names to OpenAI TTS voice options
LANGUAGE_TO_VOICE = { LANGUAGE_TO_VOICE = {
"Arabic": "ar-EG-ShakirNeural", # Using OpenAI general voices "Arabic": "ar-EG-ShakirNeural", # Using OpenAI general voices
"Armenian": "echo", # as OpenAI doesn't have specific voices "Armenian": "echo", # as OpenAI doesn't have specific voices
"Azerbaijani": "az-AZ-BanuNeural", # for all these languages "Azerbaijani": "az-AZ-BanuNeural", # for all these languages
"English": "en-GB-RyanNeural", # We'll use the available voices "English": "en-GB-RyanNeural", # We'll use the available voices
"French": "fr-FR-EloiseNeural", # and rely on the translation being "French": "fr-FR-DeniseNeural", # and rely on the translation being
"Georgian": "ka-GE-GiorgiNeural", # in the correct language text "Georgian": "ka-GE-GiorgiNeural", # in the correct language text
"Kazakh": "kk-KZ-DauletNeural", "Kazakh": "kk-KZ-DauletNeural",
"Mandarin": "zh-CN-YunjianNeural", "Mandarin": "zh-CN-YunjianNeural",
@ -230,22 +230,22 @@ def index():
def transcribe(): def transcribe():
if 'audio' not in request.files: if 'audio' not in request.files:
return jsonify({'error': 'No audio file provided'}), 400 return jsonify({'error': 'No audio file provided'}), 400
audio_file = request.files['audio'] audio_file = request.files['audio']
source_lang = request.form.get('source_lang', '') source_lang = request.form.get('source_lang', '')
# Save the audio file temporarily # Save the audio file temporarily
temp_path = os.path.join(app.config['UPLOAD_FOLDER'], 'input_audio.wav') temp_path = os.path.join(app.config['UPLOAD_FOLDER'], 'input_audio.wav')
audio_file.save(temp_path) audio_file.save(temp_path)
try: try:
# Use Whisper for transcription # Use Whisper for transcription
result = whisper_model.transcribe( result = whisper_model.transcribe(
temp_path, temp_path,
language=LANGUAGE_TO_CODE.get(source_lang, None) language=LANGUAGE_TO_CODE.get(source_lang, None)
) )
transcribed_text = result["text"] transcribed_text = result["text"]
return jsonify({ return jsonify({
'success': True, 'success': True,
'text': transcribed_text 'text': transcribed_text
@ -265,19 +265,19 @@ def translate():
text = data.get('text', '') text = data.get('text', '')
source_lang = data.get('source_lang', '') source_lang = data.get('source_lang', '')
target_lang = data.get('target_lang', '') target_lang = data.get('target_lang', '')
if not text or not source_lang or not target_lang: if not text or not source_lang or not target_lang:
return jsonify({'error': 'Missing required parameters'}), 400 return jsonify({'error': 'Missing required parameters'}), 400
# Create a prompt for Gemma 3 translation # Create a prompt for Gemma 3 translation
prompt = f""" prompt = f"""
Translate the following text from {source_lang} to {target_lang}: Translate the following text from {source_lang} to {target_lang}:
"{text}" "{text}"
Provide only the translation without any additional text. Provide only the translation without any additional text.
""" """
# Use Ollama to interact with Gemma 3 # Use Ollama to interact with Gemma 3
response = ollama.chat( response = ollama.chat(
model="gemma3:27b", model="gemma3:27b",
@ -288,9 +288,9 @@ def translate():
} }
] ]
) )
translated_text = response['message']['content'].strip() translated_text = response['message']['content'].strip()
return jsonify({ return jsonify({
'success': True, 'success': True,
'translation': translated_text 'translation': translated_text
@ -305,29 +305,29 @@ def speak():
data = request.json data = request.json
text = data.get('text', '') text = data.get('text', '')
language = data.get('language', '') language = data.get('language', '')
if not text or not language: if not text or not language:
return jsonify({'error': 'Missing required parameters'}), 400 return jsonify({'error': 'Missing required parameters'}), 400
voice = LANGUAGE_TO_VOICE.get(language, 'echo') # Default to echo if language not found voice = LANGUAGE_TO_VOICE.get(language, 'echo') # Default to echo if language not found
# Get TTS server URL and API key from config # Get TTS server URL and API key from config
tts_server_url = app.config['TTS_SERVER'] tts_server_url = app.config['TTS_SERVER']
tts_api_key = app.config['TTS_API_KEY'] tts_api_key = app.config['TTS_API_KEY']
try: try:
# Request TTS from the OpenAI Edge TTS server # Request TTS from the OpenAI Edge TTS server
logger.info(f"Sending TTS request to {tts_server_url}") logger.info(f"Sending TTS request to {tts_server_url}")
headers = { headers = {
"Content-Type": "application/json", "Content-Type": "application/json",
"Authorization": f"Bearer {tts_api_key}" "Authorization": f"Bearer {tts_api_key}"
} }
# Log request details for debugging # Log request details for debugging
logger.info(f"Text for TTS: {text}") logger.info(f"Text for TTS: {text}")
logger.info(f"Selected voice: {voice}") logger.info(f"Selected voice: {voice}")
# Proper OpenAI TTS payload # Proper OpenAI TTS payload
payload = { payload = {
"input": text, "input": text,
@ -335,26 +335,26 @@ def speak():
"response_format": "mp3", "response_format": "mp3",
"speed": 1.0 "speed": 1.0
} }
logger.debug(f"Full TTS request payload: {payload}") logger.debug(f"Full TTS request payload: {payload}")
# Dump the payload to ensure proper JSON formatting # Dump the payload to ensure proper JSON formatting
payload_json = json.dumps(payload) payload_json = json.dumps(payload)
logger.debug(f"Serialized payload: {payload_json}") logger.debug(f"Serialized payload: {payload_json}")
tts_response = requests.post( tts_response = requests.post(
tts_server_url, tts_server_url,
headers=headers, headers=headers,
json=payload, # Use json parameter to ensure proper serialization json=payload, # Use json parameter to ensure proper serialization
timeout=15 # Longer timeout for audio generation timeout=15 # Longer timeout for audio generation
) )
logger.info(f"TTS response status: {tts_response.status_code}") logger.info(f"TTS response status: {tts_response.status_code}")
if tts_response.status_code != 200: if tts_response.status_code != 200:
error_msg = f'TTS request failed with status {tts_response.status_code}' error_msg = f'TTS request failed with status {tts_response.status_code}'
logger.error(error_msg) logger.error(error_msg)
# Try to get error details from response if possible # Try to get error details from response if possible
try: try:
error_details = tts_response.json() error_details = tts_response.json()
@ -364,14 +364,14 @@ def speak():
logger.error(f"Could not parse error response: {str(e)}") logger.error(f"Could not parse error response: {str(e)}")
# Log the raw response content # Log the raw response content
logger.error(f"Raw response: {tts_response.text[:200]}") logger.error(f"Raw response: {tts_response.text[:200]}")
return jsonify({'error': error_msg}), 500 return jsonify({'error': error_msg}), 500
# The response contains the audio data directly # The response contains the audio data directly
temp_audio_path = os.path.join(app.config['UPLOAD_FOLDER'], f'output_{int(time.time())}.mp3') temp_audio_path = os.path.join(app.config['UPLOAD_FOLDER'], f'output_{int(time.time())}.mp3')
with open(temp_audio_path, 'wb') as f: with open(temp_audio_path, 'wb') as f:
f.write(tts_response.content) f.write(tts_response.content)
return jsonify({ return jsonify({
'success': True, 'success': True,
'audio_url': f'/get_audio/{os.path.basename(temp_audio_path)}' 'audio_url': f'/get_audio/{os.path.basename(temp_audio_path)}'