killua/TakwayPlatform
1
0 Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat: 后端封装tts，可以通过配置文件切换openvoice和vits

Browse Source
This commit is contained in:
killua committed 2024-05-24 15:08:55 +08:00
1 parent e2f3decfae
commit ce033dca2b
6 files changed
+49 -62
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

app/controllers/chat_controller.py
+7 -35
View File
@@ -11,6 +11,7 @@ from datetime import datetime
from utils.xf_asr_utils import generate_xf_asr_url
from config import get_config
import numpy as np
import struct
import uuid
import json
import asyncio
@@ -319,20 +320,11 @@ async def sct_llm_handler(ws,session_id,response_type,llm_info,tts_info,db,redis
response_message = {"type": "text", "code":200, "msg": sentence}
await ws.send_text(json.dumps(response_message, ensure_ascii=False)) #返回文本信息
elif response_type == RESPONSE_AUDIO:
if target_se.size == 0:
audio,sr = tts._base_tts(text=sentence,
noise_scale=tts_info["noise_scale"],
noise_scale_w=tts_info["noise_scale_w"],
speed=tts_info["speed"])
else:
audio,sr = tts.synthesize(text=sentence,
noise_scale=tts_info["noise_scale"],
noise_scale_w=tts_info["noise_scale_w"],
speed=tts_info["speed"],
target_se=target_se)
audio,sr = tts.synthesize(text=sentence,tts_info=tts_info,target_se=target_se)
response_message = {"type": "text", "code":200, "msg": sentence}
await ws.send_bytes(audio) #返回音频数据
await ws.send_text(json.dumps(response_message, ensure_ascii=False)) #返回文本信息
response_bytes = json.dumps(response_message, ensure_ascii=False).encode('utf-8')
header = struct.pack('!II',len(response_bytes),len(audio))
message_bytes = header + response_bytes + audio
logger.debug(f"websocket返回: {sentence}")
if is_end:
logger.debug(f"llm返回结果: {llm_response}")
@@ -500,17 +492,7 @@ async def scl_llm_handler(ws,session_id,response_type,llm_info,tts_info,db,redis
response_message = {"type": "text", "code":200, "msg": sentence}
await ws.send_text(json.dumps(response_message, ensure_ascii=False))
elif response_type == RESPONSE_AUDIO:
if target_se.size == 0:
audio,sr = tts._base_tts(text=sentence,
noise_scale=tts_info["noise_scale"],
noise_scale_w=tts_info["noise_scale_w"],
speed=tts_info["speed"])
else:
audio,sr = tts.synthesize(text=sentence,
noise_scale=tts_info["noise_scale"],
noise_scale_w=tts_info["noise_scale_w"],
speed=tts_info["speed"],
target_se=target_se)
audio,sr = tts.synthesize(text=sentence,tts_info=tts_info,target_se=target_se)
response_message = {"type": "text", "code":200, "msg": sentence}
await ws.send_bytes(audio)
await ws.send_text(json.dumps(response_message, ensure_ascii=False))
@@ -684,17 +666,7 @@ async def voice_call_llm_handler(ws,session_id,llm_info,tts_info,db,redis,asr_re
llm_response += chunk_data
sentences,current_sentence,is_first = split_string_with_punctuation(current_sentence,chunk_data,is_first,is_end)
for sentence in sentences:
if target_se.size == 0:
audio,sr = tts._base_tts(text=sentence,
noise_scale=tts_info["noise_scale"],
noise_scale_w=tts_info["noise_scale_w"],
speed=tts_info["speed"])
else:
audio,sr = tts.synthesize(text=sentence,
noise_scale=tts_info["noise_scale"],
noise_scale_w=tts_info["noise_scale_w"],
speed=tts_info["speed"],
target_se=target_se)
audio,sr = tts.synthesize(text=sentence,tts_info=tts_info,target_se=target_se)
text_response = {"type": "llm_text", "code": 200, "msg": sentence}
await ws.send_bytes(audio) #返回音频二进制流数据
await ws.send_text(json.dumps(text_response, ensure_ascii=False)) #返回文本数据