Files
xiaozhi-esp32-server/core/handle/sendAudioHandle.py
T

96 lines
3.0 KiB
Python
Raw Normal View History

2025-02-18 00:07:19 +08:00
from config.logger import setup_logging
2025-02-02 23:01:14 +08:00
import json
import asyncio
import time
from core.utils.util import remove_punctuation_and_length, get_string_no_punctuation_or_emoji
2025-02-18 00:07:19 +08:00
TAG = __name__
logger = setup_logging()
2025-02-02 23:01:14 +08:00
2025-02-14 23:09:12 +08:00
async def isLLMWantToFinish(conn):
first_text = conn.tts_first_text
last_text = conn.tts_last_text
_, last_text_without_punctuation = remove_punctuation_and_length(last_text)
if "再见" in last_text_without_punctuation or "拜拜" in last_text_without_punctuation:
return True
_, first_text_without_punctuation = remove_punctuation_and_length(first_text)
if "再见" in first_text_without_punctuation or "拜拜" in first_text_without_punctuation:
return True
return False
2025-02-02 23:01:14 +08:00
async def sendAudioMessage(conn, audios, duration, text):
base_delay = conn.tts_duration
# 发送 tts.start
2025-02-02 23:01:14 +08:00
if text == conn.tts_first_text:
2025-02-18 00:07:19 +08:00
logger.bind(tag=TAG).info(f"发送第一段语音: {text}")
2025-02-02 23:01:14 +08:00
conn.tts_start_speak_time = time.time()
# 发送 sentence_start(每个音频文件之前发送一次)
sentence_task = asyncio.create_task(
schedule_with_interrupt(base_delay, send_tts_message(conn, "sentence_start", text))
)
conn.scheduled_tasks.append(sentence_task)
conn.tts_duration += duration
2025-02-02 23:01:14 +08:00
# 发送音频数据
for idx, opus_packet in enumerate(audios):
2025-02-02 23:01:14 +08:00
await conn.websocket.send(opus_packet)
if conn.llm_finish_task and text == conn.tts_last_text:
2025-02-11 13:51:44 +08:00
stop_duration = conn.tts_duration - (time.time() - conn.tts_start_speak_time)
2025-02-02 23:01:14 +08:00
stop_task = asyncio.create_task(
2025-02-11 13:51:44 +08:00
schedule_with_interrupt(stop_duration, send_tts_message(conn, 'stop'))
2025-02-02 23:01:14 +08:00
)
conn.scheduled_tasks.append(stop_task)
2025-02-14 23:09:12 +08:00
if await isLLMWantToFinish(conn):
finish_task = asyncio.create_task(
2025-02-26 01:33:05 +08:00
schedule_with_interrupt(stop_duration, await conn.close())
2025-02-14 23:09:12 +08:00
)
conn.scheduled_tasks.append(finish_task)
2025-02-02 23:01:14 +08:00
async def send_tts_message(conn, state, text=None):
"""发送 TTS 状态消息"""
message = {
2025-02-02 23:01:14 +08:00
"type": "tts",
"state": state,
2025-02-02 23:01:14 +08:00
"session_id": conn.session_id
}
if text is not None:
message["text"] = text
2025-02-02 23:01:14 +08:00
await conn.websocket.send(json.dumps(message))
if state == "stop":
conn.clearSpeakStatus()
2025-02-02 23:01:14 +08:00
2025-02-11 13:51:44 +08:00
async def send_stt_message(conn, text):
"""发送 STT 状态消息"""
stt_text = get_string_no_punctuation_or_emoji(text)
await conn.websocket.send(json.dumps({
"type": "stt",
"text": stt_text,
"session_id": conn.session_id}
))
await conn.websocket.send(
json.dumps({
"type": "llm",
"text": "😊",
"emotion": "happy",
"session_id": conn.session_id}
))
await send_tts_message(conn, "start")
2025-02-02 23:01:14 +08:00
async def schedule_with_interrupt(delay, coro):
"""可中断的延迟调度"""
try:
await asyncio.sleep(delay)
await coro
except asyncio.CancelledError:
2025-02-15 11:48:02 +08:00
pass