Skip to content

Commit

Permalink
TTS中的vits、bert-vits2、gpt-sovits的float类型的相关参数支持范围语法
Browse files Browse the repository at this point in the history
  • Loading branch information
@Ikaros-521
Ikaros-521 committed Mar 13, 2024
1 parent 344006f commit 4373aa4
Show file tree
Hide file tree
Showing 4 changed files with 34 additions and 17 deletions.
2 changes: 1 addition & 1 deletion README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@

`Luna AI` 的外观由 `Live2D、Vtube Studio、xuniren、UE5 结合 Audio2Face、EasyAIVtuber、数字人视频播放器（Easy-Wav2Lip）` 技术打造，为用户提供了一个生动、互动的虚拟形象。这使得 `Luna AI` 能够在各大直播平台，如 `Bilibili、抖音、快手、微信视频号、斗鱼、YouTube、Twitch 和 TikTok`，进行实时互动直播。当然，它也可以在本地环境中与您进行个性化对话。

为了使交流更加自然，`Luna AI` 使用了先进的自然语言处理技术，结合文本转语音系统，如 `Edge-TTS、VITS-Fast、elevenlabs、bark-gui、VALL-E-X、睿声AI、genshinvoice.top、tts.ai-lab.top、OpenVoice、GPT_SoVITS、clone-voice、Azure TTS`。这不仅让它能够生成流畅的回答，还可以通过 `so-vits-svc 和 DDSP-SVC` 实现声音的变化，以适应不同的场景和角色。
为了使交流更加自然，`Luna AI` 使用了先进的自然语言处理技术，结合文本转语音系统，如 `Edge-TTS、VITS-Fast、elevenlabs、bark-gui、VALL-E-X、睿声AI、genshinvoice.top、tts.ai-lab.top、OpenVoice、GPT_SoVITS、clone-voice、Azure TTS、fish-speech`。这不仅让它能够生成流畅的回答，还可以通过 `so-vits-svc 和 DDSP-SVC` 实现声音的变化，以适应不同的场景和角色。

此外，`Luna AI` 还能够通过特定指令与 `Stable Diffusion` 协作，展示画作。用户还可以自定义文案，让 Luna AI 循环播放，以满足不同场合的需求。

Expand Down
4 changes: 4 additions & 0 deletions main.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1256,12 +1256,16 @@ def _on_open_live_danmaku(self, client: blivedm.OpenLiveClient, message: open_mo
# logging.info(f'[{client.room_id}] {message.uname}：{message.msg}')
content = message.msg # 获取弹幕内容
username = message.uname # 获取发送弹幕的用户昵称
user_face = message.uface

logging.debug(f"用户：{username} 头像：{user_face}")

logging.info(f"[{username}]: {content}")

data = {
"platform": platform,
"username": username,
"uface": user_face,
"content": content
}

Expand Down
34 changes: 24 additions & 10 deletions utils/audio_handle/my_tts.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@
import traceback
import edge_tts
from urllib.parse import urljoin
import random

from utils.common import Common
from utils.logger import Configure_logger
Expand Down Expand Up @@ -43,6 +44,17 @@ def __init__(self, config_path):
logging.error("请检查播放音频的音频输出路径配置！！！这将影响程序使用！")


# 获取随机数，单数据就是原数值，有-则判断为范围性数据，随机一个数值，返回float数据
def get_random_float(self, data):
# 将非字符串的情况统一处理为长度相同的最小值和最大值
if isinstance(data, str) and "-" in data:
min, max = map(float, data.split("-"))
else:
min = max = float(data)

# 返回指定范围内的随机浮点数
return random.uniform(min, max)

# 请求vits的api
async def vits_api(self, data):
try:
Expand Down Expand Up @@ -76,16 +88,17 @@ async def vits_api(self, data):
elif data["type"] == "bert_vits2":
# API地址 "http://127.0.0.1:23456/voice/bert-vits2"
API_URL = urljoin(data["api_ip_port"], '/voice/bert-vits2')

data_json = {
"text": data["content"],
"id": data["id"],
"format": data["format"],
"lang": "ja",
"length": data["length"],
"noise": data["noise"],
"noisew": data["noisew"],
"length": self.get_random_float(data["length"]),
"noise": self.get_random_float(data["noise"]),
"noisew": self.get_random_float(data["noisew"]),
"max": data["max"],
"sdp_radio": data["sdp_radio"]
"sdp_radio": self.get_random_float(data["sdp_radio"])
}

if data["lang"] == "中文" or data["lang"] == "汉语":
Expand Down Expand Up @@ -141,18 +154,18 @@ async def bert_vits2_api(self, data):
"speaker_name": data["speaker_name"],
"speaker_id": data["speaker_id"],
"language": data["language"],
"length": data["length"],
"noise": data["noise"],
"noisew": data["noisew"],
"sdp_radio": data["sdp_radio"],
"length": self.get_random_float(data["length"]),
"noise": self.get_random_float(data["noise"]),
"noisew": self.get_random_float(data["noisew"]),
"sdp_radio": self.get_random_float(data["sdp_radio"]),
"auto_translate": data["auto_translate"],
"auto_split": data["auto_split"],
"emotion": data["emotion"],
"style_text": data["style_text"],
"style_weight": data["style_weight"]
"style_weight": self.get_random_float(data["style_weight"])
}

# logging.info(f"data_json={data_json}")
logging.debug(f"data_json={data_json}")
# logging.info(f"data={data}")

logging.debug(f"API_URL={API_URL}")
Expand Down Expand Up @@ -642,6 +655,7 @@ async def websocket_client_logic(websocket, data_json):
if key != "api_ip_port"
}

params["speed"] = self.get_random_float(params["speed"])
params["text"] = data["content"]

async with aiohttp.ClientSession() as session:
Expand Down
11 changes: 5 additions & 6 deletions webui.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1041,7 +1041,6 @@ def common_textarea_handle(content):
config_data["claude"]["slack_user_token"] = input_claude_slack_user_token.value
config_data["claude"]["bot_user_id"] = input_claude_bot_user_id.value

if config.get("webui", "show_card", "llm", "claude2"):
config_data["claude2"]["cookie"] = input_claude2_cookie.value
config_data["claude2"]["use_proxy"] = switch_claude2_use_proxy.value
config_data["claude2"]["proxies"]["http"] = input_claude2_proxies_http.value
Expand Down Expand Up @@ -1277,13 +1276,13 @@ def common_textarea_handle(content):
config_data["bert_vits2"]["speaker_name"] = input_vits_speaker_name.value
config_data["bert_vits2"]["speaker_id"] = int(input_vits_speaker_id.value)
config_data["bert_vits2"]["language"] = select_bert_vits2_language.value
config_data["bert_vits2"]["length"] = round(float(input_bert_vits2_length.value), 2)
config_data["bert_vits2"]["noise"] = round(float(input_bert_vits2_noise.value), 2)
config_data["bert_vits2"]["noisew"] = round(float(input_bert_vits2_noisew.value), 2)
config_data["bert_vits2"]["sdp_radio"] = round(float(input_bert_vits2_sdp_radio.value), 2)
config_data["bert_vits2"]["length"] = input_bert_vits2_length.value
config_data["bert_vits2"]["noise"] = input_bert_vits2_noise.value
config_data["bert_vits2"]["noisew"] = input_bert_vits2_noisew.value
config_data["bert_vits2"]["sdp_radio"] = input_bert_vits2_sdp_radio.value
config_data["bert_vits2"]["emotion"] = input_bert_vits2_emotion.value
config_data["bert_vits2"]["style_text"] = input_bert_vits2_style_text.value
config_data["bert_vits2"]["style_weight"] = round(float(input_bert_vits2_style_weight.value), 2)
config_data["bert_vits2"]["style_weight"] = input_bert_vits2_style_weight.value
config_data["bert_vits2"]["auto_translate"] = switch_bert_vits2_auto_translate.value
config_data["bert_vits2"]["auto_split"] = switch_bert_vits2_auto_split.value

Expand Down

0 comments on commit 4373aa4

Please sign in to comment.