Spaces:
Sleeping
Sleeping
File size: 4,796 Bytes
8c3fde1 e067d0c 8c3fde1 e067d0c 8c3fde1 e067d0c 8c3fde1 e067d0c |
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 |
from flask import Flask, request, jsonify, render_template
from flask_cors import CORS
from faster_whisper import WhisperModel
from opencc import OpenCC
import os
import tempfile
from werkzeug.utils import secure_filename
app = Flask(__name__)
CORS(app)
# 配置
UPLOAD_FOLDER = 'uploads'
ALLOWED_EXTENSIONS = {'mp3', 'wav', 'ogg', 'flac', 'm4a', 'webm'}
MAX_FILE_SIZE = 100 * 1024 * 1024 # 100MB
os.makedirs(UPLOAD_FOLDER, exist_ok=True)
# 全局模型變數
model = None
# OpenCC 轉換器（簡體轉繁體台灣）
cc_s2tw = OpenCC('s2tw') # 簡體到繁體（台灣）
cc_s2twp = OpenCC('s2twp') # 簡體到繁體（台灣）含常用詞彙轉換
def convert_to_traditional(text, use_phrases=True):
"""將簡體中文轉換為繁體中文（台灣）"""
if use_phrases:
return cc_s2twp.convert(text)
return cc_s2tw.convert(text)
def allowed_file(filename):
return '.' in filename and filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS
def load_model():
"""延迟加载模型"""
global model
if model is None:
print("正在加载 Whisper 模型（CPU 模式）...")
# 明確指定使用 CPU，compute_type="int8" 適合 CPU 運算
model = WhisperModel(
"dropbox-dash/faster-whisper-large-v3-turbo",
device="cpu",
compute_type="int8",
cpu_threads=4, # 限制 CPU 線程數，避免佔用過多資源
num_workers=1 # 減少並行工作數
)
print("模型加载完成！")
return model
@app.route('/')
def index():
return render_template('index.html')
@app.route('/api/transcribe', methods=['POST'])
def transcribe():
try:
# 檢查是否有檔案
if 'audio' not in request.files:
return jsonify({'error': '沒有上傳檔案'}), 400
file = request.files['audio']
if file.filename == '':
return jsonify({'error': '沒有選擇檔案'}), 400
if not allowed_file(file.filename):
return jsonify({'error': f'不支援的檔案格式。支援的格式: {", ".join(ALLOWED_EXTENSIONS)}'}), 400
# 獲取參數
language = request.form.get('language', None)
beam_size = int(request.form.get('beam_size', 5))
to_traditional = request.form.get('to_traditional', 'true').lower() == 'true'
# 儲存臨時檔案
filename = secure_filename(file.filename)
filepath = os.path.join(UPLOAD_FOLDER, filename)
file.save(filepath)
try:
# 載入模型並轉錄
whisper_model = load_model()
# 使用 initial_prompt 引導輸出繁體中文
initial_prompt = "以下是普通話的轉錄內容。" if language == 'zh' else None
segments, info = whisper_model.transcribe(
filepath,
language=language if language else None,
beam_size=beam_size,
vad_filter=True,
initial_prompt=initial_prompt
)
# 收集結果
results = []
full_text = ""
for segment in segments:
text = segment.text.strip()
# 如果是中文且啟用繁體轉換
if to_traditional and info.language == 'zh':
text = convert_to_traditional(text)
segment_data = {
'start': round(segment.start, 2),
'end': round(segment.end, 2),
'text': text
}
results.append(segment_data)
full_text += text + " "
# 刪除臨時檔案
os.remove(filepath)
return jsonify({
'success': True,
'language': info.language,
'duration': round(info.duration, 2),
'full_text': full_text.strip(),
'segments': results
})
except Exception as e:
# 清理臨時檔案
if os.path.exists(filepath):
os.remove(filepath)
raise e
except Exception as e:
return jsonify({'error': f'轉錄失敗: {str(e)}'}), 500
@app.route('/api/health', methods=['GET'])
def health():
return jsonify({'status': 'ok', 'model_loaded': model is not None})
if __name__ == '__main__':
# 從環境變數讀取端口，預設為 7860（Hugging Face Spaces）
port = int(os.environ.get('PORT', 7860))
print("=" * 50)
print("Whisper 語音轉文字服務")
print("=" * 50)
print(f"服務將在 http://0.0.0.0:{port} 啟動")
print("首次轉錄時會自動下載模型，請耐心等待...")
print("=" * 50)
app.run(debug=False, host='0.0.0.0', port=port)
|