-
Notifications
You must be signed in to change notification settings - Fork 10
Expand file tree
/
Copy pathexample_simple_usage.py
More file actions
134 lines (103 loc) · 4.29 KB
/
Copy pathexample_simple_usage.py
File metadata and controls
134 lines (103 loc) · 4.29 KB
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
#!/usr/bin/env python3
"""
简单的VAD测试脚本
测试重构后的Cascade StreamProcessor，使用两个.wav文件进行流式VAD检测，
并将检测到的语音段保存为独立的.wav文件。
"""
import asyncio
import os
import wave
from pathlib import Path
import cascade
async def save_speech_segment(audio_data: bytes, output_path: str):
"""
保存语音段为WAV文件
Args:
audio_data: 音频数据（16kHz, 16bit, mono）
output_path: 输出文件路径
"""
try:
with wave.open(output_path, 'wb') as wav_file:
wav_file.setnchannels(1) # 单声道
wav_file.setsampwidth(2) # 16位
wav_file.setframerate(16000) # 16kHz采样率
wav_file.writeframes(audio_data)
print(f"💾 已保存语音段: {output_path}")
except Exception as e:
print(f"❌ 保存失败 {output_path}: {e}")
async def test_vad_on_file(audio_file: str):
"""
测试单个音频文件的VAD处理
Args:
audio_file: 音频文件路径
"""
if not os.path.exists(audio_file):
print(f"❌ 文件不存在: {audio_file}")
return
print(f"\n🎯 开始处理: {Path(audio_file).name}")
print("=" * 50)
# 创建输出目录
output_dir = Path("speech_segments")
output_dir.mkdir(exist_ok=True)
# 文件名前缀
file_prefix = Path(audio_file).stem
segment_count = 0
frame_count = 0
try:
# 创建配置
config = cascade.Config(
vad_threshold=0.5,
min_silence_duration_ms=500,
speech_pad_ms=300
)
# 使用StreamProcessor处理文件
async with cascade.StreamProcessor(config) as processor:
print("✅ StreamProcessor已启动（独立VAD模型）")
async for result in processor.process_file(audio_file):
if result.is_speech_segment and result.segment:
segment_count += 1
segment = result.segment
# 生成输出文件名
start_ms = int(segment.start_timestamp_ms)
end_ms = int(segment.end_timestamp_ms)
duration_ms = int(segment.duration_ms)
output_filename = f"{file_prefix}_segment_{segment_count:03d}_{start_ms}ms-{end_ms}ms.wav"
output_path = output_dir / output_filename
# 保存语音段
await save_speech_segment(segment.audio_data, str(output_path))
print(f"🎤 语音段 {segment_count}: {start_ms}ms - {end_ms}ms "
f"(时长: {duration_ms}ms, {segment.frame_count}帧)")
elif result.frame:
frame_count += 1
if frame_count % 50 == 0: # 每50帧打印一次
print(f"🔇 处理帧: {frame_count}", end="\r")
# 获取统计信息
stats = processor.get_stats()
print(f"\n📊 处理完成:")
print(f" 🎤 语音段数量: {segment_count}")
print(f" 🔇 单帧数量: {frame_count}")
print(f" 📦 总处理块: {stats.total_chunks_processed}")
print(f" ⏱️ 平均处理时间: {stats.average_processing_time_ms:.2f}ms")
print(f" 💾 语音段保存到: {output_dir.absolute()}")
except Exception as e:
print(f"❌ 处理失败: {e}")
async def main():
"""主函数"""
print("🌊 Cascade 简单VAD测试")
print("基于重构后的1:1:1:1架构")
print("=" * 50)
# 测试文件列表
test_files = [
"我现在开始录音，理论上会有两个文件.wav"
]
# 检查并处理每个文件
for audio_file in test_files:
if os.path.exists(audio_file):
await test_vad_on_file(audio_file)
else:
print(f"⚠️ 跳过不存在的文件: {audio_file}")
print(f"\n🎉 测试完成！")
print(f"✅ 重构后的StreamProcessor工作正常")
print(f"✅ 独立模型架构无并发问题")
if __name__ == "__main__":
asyncio.run(main())