Spaces:
Building
Building
File size: 8,685 Bytes
edec17e |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 |
"""
Audio Buffer Manager for Flare
==============================
Manages audio buffering, silence detection, and chunk processing
"""
import asyncio
from typing import Dict, Optional, List, Tuple, Any
from collections import deque
from datetime import datetime
import base64
import numpy as np
from dataclasses import dataclass
import traceback
from chat_session.event_bus import EventBus, Event, EventType
from utils.logger import log_info, log_error, log_debug, log_warning
@dataclass
class AudioChunk:
"""Audio chunk with metadata"""
data: bytes
timestamp: datetime
chunk_index: int
is_speech: bool = True
energy_level: float = 0.0
class SilenceDetector:
"""Detect silence in audio stream"""
def __init__(self,
threshold_ms: int = 2000,
energy_threshold: float = 0.01,
sample_rate: int = 16000):
self.threshold_ms = threshold_ms
self.energy_threshold = energy_threshold
self.sample_rate = sample_rate
self.silence_start: Optional[datetime] = None
def detect_silence(self, audio_chunk: bytes) -> Tuple[bool, int]:
"""
Detect if chunk is silence and return duration
Returns: (is_silence, silence_duration_ms)
"""
try:
# Handle empty or invalid chunk
if not audio_chunk or len(audio_chunk) < 2:
return True, 0
# Ensure even number of bytes for 16-bit audio
if len(audio_chunk) % 2 != 0:
audio_chunk = audio_chunk[:-1]
# Convert to numpy array
audio_data = np.frombuffer(audio_chunk, dtype=np.int16)
if len(audio_data) == 0:
return True, 0
# Calculate RMS energy
rms = np.sqrt(np.mean(audio_data.astype(float) ** 2))
normalized_rms = rms / 32768.0 # Normalize for 16-bit audio
is_silence = normalized_rms < self.energy_threshold
# Track silence duration
now = datetime.utcnow()
if is_silence:
if self.silence_start is None:
self.silence_start = now
duration_ms = int((now - self.silence_start).total_seconds() * 1000)
else:
self.silence_start = None
duration_ms = 0
return is_silence, duration_ms
except Exception as e:
log_warning(f"Silence detection error: {e}")
return False, 0
def reset(self):
"""Reset silence detection state"""
self.silence_start = None
class AudioBuffer:
"""Manage audio chunks for a session"""
def __init__(self,
session_id: str,
max_chunks: int = 1000,
chunk_size_bytes: int = 4096):
self.session_id = session_id
self.max_chunks = max_chunks
self.chunk_size_bytes = chunk_size_bytes
self.chunks: deque[AudioChunk] = deque(maxlen=max_chunks)
self.chunk_counter = 0
self.total_bytes = 0
self.lock = asyncio.Lock()
async def add_chunk(self, audio_data: bytes, timestamp: Optional[datetime] = None) -> AudioChunk:
"""Add audio chunk to buffer"""
async with self.lock:
if timestamp is None:
timestamp = datetime.utcnow()
chunk = AudioChunk(
data=audio_data,
timestamp=timestamp,
chunk_index=self.chunk_counter
)
self.chunks.append(chunk)
self.chunk_counter += 1
self.total_bytes += len(audio_data)
return chunk
async def get_recent_audio(self, duration_ms: int = 5000) -> bytes:
"""Get recent audio data"""
async with self.lock:
cutoff_time = datetime.utcnow()
audio_parts = []
# Iterate backwards through chunks
for chunk in reversed(self.chunks):
time_diff = (cutoff_time - chunk.timestamp).total_seconds() * 1000
if time_diff > duration_ms:
break
audio_parts.append(chunk.data)
# Reverse to maintain chronological order
audio_parts.reverse()
return b''.join(audio_parts)
async def clear(self):
"""Clear buffer"""
async with self.lock:
self.chunks.clear()
self.chunk_counter = 0
self.total_bytes = 0
def get_stats(self) -> Dict[str, Any]:
"""Get buffer statistics"""
return {
"chunks": len(self.chunks),
"total_bytes": self.total_bytes,
"chunk_counter": self.chunk_counter,
"oldest_chunk": self.chunks[0].timestamp if self.chunks else None,
"newest_chunk": self.chunks[-1].timestamp if self.chunks else None
}
class AudioBufferManager:
"""Manage audio buffers for all sessions"""
def __init__(self, event_bus: EventBus):
self.event_bus = event_bus
self.session_buffers: Dict[str, AudioBuffer] = {}
self.silence_detectors: Dict[str, SilenceDetector] = {}
self._setup_event_handlers()
def _setup_event_handlers(self):
"""Subscribe to audio events"""
self.event_bus.subscribe(EventType.SESSION_STARTED, self._handle_session_started)
self.event_bus.subscribe(EventType.SESSION_ENDED, self._handle_session_ended)
self.event_bus.subscribe(EventType.AUDIO_CHUNK_RECEIVED, self._handle_audio_chunk)
async def _handle_session_started(self, event: Event):
"""Initialize buffer for new session"""
session_id = event.session_id
config = event.data
# Create audio buffer
self.session_buffers[session_id] = AudioBuffer(
session_id=session_id,
max_chunks=config.get("max_chunks", 1000),
chunk_size_bytes=config.get("chunk_size", 4096)
)
log_info(f"📦 Audio buffer initialized", session_id=session_id)
async def _handle_session_ended(self, event: Event):
"""Cleanup session buffers"""
session_id = event.session_id
# Clear and remove buffer
if session_id in self.session_buffers:
await self.session_buffers[session_id].clear()
del self.session_buffers[session_id]
# Remove silence detector
if session_id in self.silence_detectors:
del self.silence_detectors[session_id]
log_info(f"📦 Audio buffer cleaned up", session_id=session_id)
async def _handle_audio_chunk(self, event: Event):
"""Process incoming audio chunk"""
session_id = event.session_id
buffer = self.session_buffers.get(session_id)
if not buffer:
log_warning(f"⚠️ No buffer for session", session_id=session_id)
return
try:
# Decode audio data
audio_data = base64.b64decode(event.data.get("audio_data", ""))
# Add to buffer
chunk = await buffer.add_chunk(audio_data)
# Log periodically
if chunk.chunk_index % 100 == 0:
stats = buffer.get_stats()
log_debug(
f"📊 Buffer stats",
session_id=session_id,
**stats
)
except Exception as e:
log_error(
f"❌ Error processing audio chunk",
session_id=session_id,
error=str(e),
traceback=traceback.format_exc()
)
async def get_buffer(self, session_id: str) -> Optional[AudioBuffer]:
"""Get buffer for session"""
return self.session_buffers.get(session_id)
async def reset_buffer(self, session_id: str):
"""Reset buffer for new utterance"""
buffer = self.session_buffers.get(session_id)
detector = self.silence_detectors.get(session_id)
if buffer:
await buffer.clear()
if detector:
detector.reset()
log_debug(f"🔄 Audio buffer reset", session_id=session_id)
def get_all_stats(self) -> Dict[str, Dict[str, Any]]:
"""Get statistics for all buffers"""
stats = {}
for session_id, buffer in self.session_buffers.items():
stats[session_id] = buffer.get_stats()
return stats |