Spaces:
Runtime error
Runtime error
File size: 3,392 Bytes
bb2885a 69067ae d445f81 2c6d73a bb2885a 56becc0 2c6d73a 56becc0 2c6d73a 56becc0 2c6d73a 56becc0 2c6d73a bb2885a 2c6d73a bb2885a 2c6d73a bb2885a 2c6d73a 56becc0 2c6d73a bb2885a 2c6d73a bb2885a 2c6d73a bb2885a 2c6d73a 56becc0 2c6d73a bb2885a 2c6d73a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 |
import torch
from flask import Flask, render_template, request, jsonify
import os
import re
import ffmpeg
from transformers import pipeline # ✅ Using correct Whisper ASR pipeline
from gtts import gTTS
from pydub import AudioSegment
from pydub.silence import detect_nonsilent
from waitress import serve
app = Flask(__name__)
# ✅ Load Whisper ASR Model correctly
device = "cuda" if torch.cuda.is_available() else "cpu"
asr_model = pipeline("automatic-speech-recognition", model="openai/whisper-large-v3", device=0 if device == "cuda" else -1)
# Function to generate audio prompts
def generate_audio_prompt(text, filename):
tts = gTTS(text=text, lang="en")
tts.save(os.path.join("static", filename))
# Generate required voice prompts
prompts = {
"welcome": "Welcome to Biryani Hub.",
"ask_name": "Tell me your name.",
"ask_email": "Please provide your email address.",
"thank_you": "Thank you for registration."
}
for key, text in prompts.items():
generate_audio_prompt(text, f"{key}.mp3")
# Symbol mapping for proper recognition
SYMBOL_MAPPING = {
"at the rate": "@",
"at": "@",
"dot": ".",
"underscore": "_",
"hash": "#",
"plus": "+",
"dash": "-",
"comma": ",",
"space": " "
}
# Function to convert audio to WAV format
def convert_to_wav(input_path, output_path):
try:
audio = AudioSegment.from_file(input_path)
audio.export(output_path, format="wav")
except Exception as e:
raise Exception(f"Audio conversion failed: {str(e)}")
# Function to clean transcribed text
def clean_transcription(text):
text = text.lower().strip()
ignore_phrases = ["my name is", "this is", "i am", "it's", "name"]
for phrase in ignore_phrases:
text = text.replace(phrase, "").strip()
for word, symbol in SYMBOL_MAPPING.items():
text = text.replace(word, symbol)
return text.capitalize()
# Function to check if audio contains actual speech
def is_silent_audio(audio_path):
audio = AudioSegment.from_wav(audio_path)
nonsilent_parts = detect_nonsilent(audio, min_silence_len=500, silence_thresh=audio.dBFS-16)
return len(nonsilent_parts) == 0 # Returns True if silence detected
@app.route("/")
def index():
return render_template("index.html")
@app.route("/transcribe", methods=["POST"])
def transcribe():
if "audio" not in request.files:
return jsonify({"error": "No audio file provided"}), 400
audio_file = request.files["audio"]
input_audio_path = os.path.join("static", "temp_input.wav")
output_audio_path = os.path.join("static", "temp.wav")
audio_file.save(input_audio_path)
try:
# Convert to WAV
convert_to_wav(input_audio_path, output_audio_path)
# Check for silence
if is_silent_audio(output_audio_path):
return jsonify({"error": "No speech detected. Please try again."}), 400
# ✅ Use Whisper ASR model for transcription
result = asr_model(output_audio_path, generate_kwargs={"language": "en"})
transcribed_text = clean_transcription(result["text"])
return jsonify({"text": transcribed_text})
except Exception as e:
return jsonify({"error": f"Speech recognition error: {str(e)}"}), 500
# Start Waitress Production Server
if __name__ == "__main__":
serve(app, host="0.0.0.0", port=7860)
|