File size: 6,899 Bytes
69067ae
d445f81
2c6d73a
 
69067ae
2c6d73a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4b54307
3d6f557
0197ed3
adb5e2a
d445f81
 
 
4b54307
0197ed3
69067ae
 
4b54307
7494646
4b54307
7494646
7467739
 
69067ae
7467739
 
4b54307
7467739
 
 
 
 
 
adb5e2a
7467739
 
685e8d2
3d6f557
78a5c3d
 
e215f5e
78a5c3d
 
 
 
 
 
 
 
 
3d6f557
9fcd178
 
 
 
 
 
 
3d6f557
 
4b54307
 
 
 
3d6f557
78a5c3d
 
3d6f557
 
d445f81
5207ead
 
d445f81
4b54307
5207ead
a3e60d6
69067ae
7467739
69067ae
 
7467739
 
69067ae
7467739
8ab530a
69067ae
9fcd178
 
 
8ab530a
7494646
3d6f557
9fcd178
 
4b54307
9fcd178
5207ead
d445f81
4b54307
9fcd178
7467739
d445f81
7467739
7494646
5207ead
0197ed3
4b54307
69067ae
36d7045
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
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
from flask import Flask, render_template, request, jsonify
import os
import torchafrom flask import Flask, render_template, request, jsonify
import os
import torch
import re
import ffmpeg  # Ensure FFmpeg is installed
from transformers import pipeline
from gtts import gTTS
from pydub import AudioSegment
from pydub.silence import detect_nonsilent
from waitress import serve
import whisper  # Improved Whisper ASR Model

app = Flask(__name__)

# Load Whisper Model for Highly Accurate Speech-to-Text
device = "cuda" if torch.cuda.is_available() else "cpu"
asr_model = whisper.load_model("large-v3", device=device)

# 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 better recognition
SYMBOL_MAPPING = {
    "at the rate": "@",
    "at": "@",
    "dot": ".",
    "underscore": "_",
    "hash": "#",
    "plus": "+",
    "dash": "-",
    "comma": ",",
    "space": " "
}

# Function to convert audio to WAV format (Fixes FFmpeg issues)
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 (Removes unnecessary words)
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 the 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")
    output_audio_path = os.path.join("static", "temp.wav")
    audio_file.save(input_audio_path)

    try:
        # Convert audio to WAV format
        convert_to_wav(input_audio_path, output_audio_path)

        # Check if the audio contains valid speech
        if is_silent_audio(output_audio_path):
            return jsonify({"error": "No speech detected. Please try again."}), 400
        
        # Transcribe using Whisper
        result = asr_model.transcribe(output_audio_path, 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

# Use Waitress for Production Server
if __name__ == "__main__":
    serve(app, host="0.0.0.0", port=7860)

import re
import ffmpeg  # Ensures FFmpeg is installed
from transformers import pipeline
from gtts import gTTS
from pydub import AudioSegment
from pydub.silence import detect_nonsilent
from waitress import serve
import whisper_timestamped  # Improved Whisper with timestamps

app = Flask(__name__)

# Load Whisper Model for Highly Accurate Speech-to-Text
device = "cuda" if torch.cuda.is_available() else "cpu"
asr_model = whisper_timestamped.load_model("medium", device=device)

# 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 (Fixes FFmpeg issues)
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 (Removes unnecessary words)
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 the 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")
    output_audio_path = os.path.join("static", "temp.wav")
    audio_file.save(input_audio_path)

    try:
        # Convert audio to WAV format
        convert_to_wav(input_audio_path, output_audio_path)

        # Check if the audio contains valid speech
        if is_silent_audio(output_audio_path):
            return jsonify({"error": "No speech detected. Please try again."}), 400
        
        # Transcribe using Whisper
        result = asr_model.transcribe(output_audio_path, 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

# Use Waitress for Production Server
if __name__ == "__main__":
    serve(app, host="0.0.0.0", port=7860)