LLM/WhisperLiveKit
1
0
Fork 0
mirror of https://github.com/QuentinFuxa/WhisperLiveKit.git synced 2026-03-07 22:33:36 +00:00
Code Issues Packages Projects Releases Wiki Activity

Batch unprocessed audio to reduce Whisper streaming calls

Browse Source
This commit is contained in:
Quentin Fuxa
2024-12-28 18:32:27 +01:00
parent 99019f1dd7
commit c960d1571d
1 changed files with 13 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
whisper_fastapi_online_server.py
View File

@@ -3,6 +3,7 @@ import argparse
import asyncio
import numpy as np
import ffmpeg
from time import time
from fastapi import FastAPI, WebSocket, WebSocketDisconnect
from fastapi.responses import HTMLResponse
@@ -69,22 +70,24 @@ async def websocket_endpoint(websocket: WebSocket):
nonlocal pcm_buffer
loop = asyncio.get_event_loop()
full_transcription = ""
beg = time()
while True:
try:
chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 4096)
if not chunk: # FFmpeg might have closed
print("FFmpeg stdout closed.")
break
elapsed_time = int(time() - beg)
beg = time()
chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 32000*elapsed_time)
if not chunk: # The first chunk will be almost empty, FFmpeg is still starting up
chunk = await loop.run_in_executor(None, ffmpeg_process.stdout.read, 4096)
if not chunk: # FFmpeg might have closed
print("FFmpeg stdout closed.")
break
pcm_buffer.extend(chunk)
while len(pcm_buffer) >= BYTES_PER_SEC:
three_sec_chunk = pcm_buffer[:BYTES_PER_SEC]
del pcm_buffer[:BYTES_PER_SEC]
if len(pcm_buffer) >= BYTES_PER_SEC:
# Convert int16 -> float32
pcm_array = np.frombuffer(three_sec_chunk, dtype=np.int16).astype(np.float32) / 32768.0
pcm_array = np.frombuffer(pcm_buffer, dtype=np.int16).astype(np.float32) / 32768.0
pcm_buffer = bytearray()
online.insert_audio_chunk(pcm_array)
transcription = online.process_iter()[2]
if args.vac: