github/NarratoAI
1
0
Fork 0
mirror of https://github.com/linyqh/NarratoAI.git synced 2026-05-01 06:08:16 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #237 from aw123456dew/feature/doubao-tts

Browse Source 
add doubao tts
This commit is contained in:
viccy 2026-04-08 15:14:10 +08:00 committed by GitHub
commit 8c129790c7
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 428 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
app/config/config.py
View File

@ -82,6 +82,7 @@ def save_config():
_cfg["ui"] = ui
_cfg["tts_qwen"] = tts_qwen
_cfg["indextts2"] = indextts2
_cfg["doubaotts"] = doubaotts
f.write(toml.dumps(_cfg))
@ -96,6 +97,7 @@ ui = _cfg.get("ui", {})
frames = _cfg.get("frames", {})
tts_qwen = _cfg.get("tts_qwen", {})
indextts2 = _cfg.get("indextts2", {})
doubaotts = _cfg.get("doubaotts", {})
hostname = socket.gethostname()

152
app/services/voice.py
View File

@ -1116,6 +1116,125 @@ def should_use_azure_speech_services(voice_name: str) -> bool:
return False
def doubaotts_tts(text: str, voice_name: str, voice_file: str, speed: float = 1.0) -> Union[SubMaker, None]:
"""
使用豆包语音 TTS 生成语音
"""
# 读取配置
doubaotts_cfg = getattr(config, "doubaotts", {}) or {}
appid = doubaotts_cfg.get("appid", "")
token = doubaotts_cfg.get("token", "")
ak = doubaotts_cfg.get("ak", "")
sk = doubaotts_cfg.get("sk", "")
cluster = doubaotts_cfg.get("cluster", "volcano_tts")
if not appid or not token:
logger.error("豆包语音 TTS 配置未完成")
return None
# 准备参数
voice_type = voice_name
safe_speed = float(max(0.2, min(3.0, speed)))
text = text.strip()
# 构建请求参数
import uuid
reqid = str(uuid.uuid4())
# 获取高级参数
volume = doubaotts_cfg.get("volume", 1.0)
pitch = doubaotts_cfg.get("pitch", 1.0)
silence_duration = doubaotts_cfg.get("silence_duration", 0.125)
payload = {
"app": {
"appid": appid,
"token": token,
"cluster": cluster
},
"user": {
"uid": "NarratoAI"
},
"audio": {
"voice_type": voice_type,
"encoding": "mp3",
"rate": 24000,
"speed_ratio": safe_speed,
"volume_ratio": float(volume),
"pitch_ratio": float(pitch)
},
"request": {
"reqid": reqid,
"text": text,
"text_type": "plain",
"operation": "query"
}
}
# 如果设置了句尾静音时长，添加到请求参数中
if silence_duration > 0:
payload["audio"]["silence_duration"] = float(silence_duration)
# API 地址
url = "https://openspeech.bytedance.com/api/v1/tts"
# 构建请求头（使用Bearer Token认证）
headers = {
"Content-Type": "application/json",
"Authorization": f"Bearer;{token}"
}
for i in range(3):
try:
logger.info(f"=== 豆包语音 TTS 请求参数 (第 {i+1} 次调用) ===")
# 发送请求
import requests
# 处理代理设置
proxies = None
proxy_enabled = config.proxy.get("enabled", False)
if proxy_enabled:
proxy_url = config.proxy.get("https", config.proxy.get("http", ""))
if proxy_url:
proxies = {"https": proxy_url, "http": proxy_url}
response = requests.post(url, json=payload, headers=headers, proxies=proxies, timeout=60)
if response.status_code == 200:
result = response.json()
if result.get("code") == 3000:
# 成功
audio_data = result.get("data", "")
if audio_data:
# 解码 base64 音频数据
import base64
audio_bytes = base64.b64decode(audio_data)
# 写入文件
with open(voice_file, "wb") as f:
f.write(audio_bytes)
logger.success(f"豆包语音 TTS 合成成功: {voice_file}")
# 创建 SubMaker 对象（简化版，不包含时间戳）
sub_maker = new_sub_maker()
return sub_maker
else:
logger.error("豆包语音 TTS 响应中无音频数据")
else:
logger.error(f"豆包语音 TTS 失败: {result.get('message', '未知错误')}")
else:
logger.error(f"豆包语音 TTS API 请求失败: {response.status_code}, {response.text}")
if i < 2:
time.sleep(1)
except Exception as e:
logger.error(f"豆包语音 TTS 错误: {str(e)}")
if i < 2:
time.sleep(3)
return None
def tts(
text: str, voice_name: str, voice_rate: float, voice_pitch: float, voice_file: str, tts_engine: str
) -> Union[SubMaker, None]:
@ -1147,6 +1266,10 @@ def tts(
if tts_engine == "indextts2":
logger.info("分发到 IndexTTS2")
return indextts2_tts(text, voice_name, voice_file, speed=voice_rate)
if tts_engine == "doubaotts":
logger.info("分发到豆包语音 TTS")
return doubaotts_tts(text, voice_name, voice_file, speed=voice_rate)
# Fallback for unknown engine - default to azure v1
logger.warning(f"未知的 TTS 引擎: '{tts_engine}', 将默认使用 Edge TTS (Azure V1)。")
@ -1606,8 +1729,8 @@ def tts_multiple(task_id: str, list_script: list, voice_name: str, voice_rate: f
f"或者使用其他 tts 引擎")
continue
else:
# SoulVoice、Qwen3、IndexTTS2 引擎不生成字幕文件
if is_soulvoice_voice(voice_name) or is_qwen_engine(tts_engine) or tts_engine == "indextts2":
# SoulVoice、Qwen3、IndexTTS2、豆包语音 引擎不生成字幕文件
if is_soulvoice_voice(voice_name) or is_qwen_engine(tts_engine) or tts_engine == "indextts2" or tts_engine == "doubaotts":
# 获取实际音频文件的时长
duration = get_audio_duration_from_file(audio_file)
if duration <= 0:
@ -1615,8 +1738,27 @@ def tts_multiple(task_id: str, list_script: list, voice_name: str, voice_rate: f
duration = get_audio_duration(sub_maker)
if duration <= 0:
# 最后的 fallback，基于文本长度估算
duration = max(1.0, len(text) / 3.0)
logger.warning(f"无法获取音频时长，使用文本估算: {duration:.2f}秒")
# 对于英文文本，使用更准确的估算方法
# 英文平均语速约为每分钟150-180个单词，即每秒2.5-3个单词
# 对于中文文本，约为每秒3-4字
import re
# 计算英文单词数
english_words = len(re.findall(r'\b\w+\b', text))
# 计算中文字符数
chinese_chars = len(re.findall(r'[\u4e00-\u9fa5]', text))
if english_words > chinese_chars:
# 主要是英文文本
# 假设平均每个单词需要0.35秒
estimated_duration = max(1.0, english_words * 0.35)
else:
# 主要是中文文本
# 假设平均每个汉字需要0.3秒
estimated_duration = max(1.0, chinese_chars * 0.3)
# 确保估算时长合理
duration = max(1.0, estimated_duration)
logger.warning(f"无法获取音频时长，使用文本估算: {duration:.2f}秒 (英文单词: {english_words}, 中文字符: {chinese_chars})")
# 不创建字幕文件
subtitle_file = ""
else:
@ -1658,8 +1800,6 @@ def get_audio_duration_from_file(audio_file: str) -> float:
# 但实际文件还包含头部信息，所以调整系数
estimated_duration = max(1.0, file_size / 20000) # 调整为更保守的估算
# 对于中文语音，根据文本长度进行二次校正
# 一般中文语音速度约为 3-4 字/秒
logger.warning(f"使用文件大小估算音频时长: {estimated_duration:.2f}秒")
return estimated_duration
except Exception as e:

22
config.example.toml
View File

@ -114,9 +114,25 @@
do_sample = true
num_beams = 3
repetition_penalty = 10.0
[doubaotts]
# 豆包语音 TTS 配置
# 申请流程：
# 1. 打开 https://console.volcengine.com/iam/keymanage 新建 Access Key 和 Secret Key
# 2. 打开 https://www.volcengine.com/product/voice-tech 点击立即使用
# 3. 在 API 服务中心找到音频生成下面的语音合成，获取 APPID 和 Token
ak = ""
sk = ""
appid = ""
token = ""
cluster = "volcano_tts"
# 高级参数
volume = 1.0
pitch = 1.0
silence_duration = 0.125
[ui]
# TTS引擎选择 (edge_tts, azure_speech, soulvoice, tencent_tts, tts_qwen)
# TTS引擎选择 (edge_tts, azure_speech, soulvoice, tencent_tts, tts_qwen, doubaotts)
tts_engine = "edge_tts"
# Edge TTS 配置
@ -130,6 +146,10 @@
azure_volume = 80
azure_rate = 1.0
azure_pitch = 0
# 豆包语音 TTS 配置
doubaotts_voice_type = "BV700_V2_streaming"
doubaotts_rate = 1.0
##########################################
# 代理和网络配置

260
webui/components/audio_settings.py
View File

@ -26,7 +26,8 @@ def get_tts_engine_options():
"azure_speech": "Azure Speech Services",
"tencent_tts": "腾讯云 TTS",
"qwen3_tts": "通义千问 Qwen3 TTS",
"indextts2": "IndexTTS2 语音克隆"
"indextts2": "IndexTTS2 语音克隆",
"doubaotts": "豆包语音 TTS"
}
@ -62,6 +63,12 @@ def get_tts_engine_descriptions():
"features": "零样本语音克隆，上传参考音频即可合成相同音色的语音，需要本地或私有部署",
"use_case": "下载地址：https://pan.quark.cn/s/0767c9bcefd5",
"registration": None
},
"doubaotts": {
"title": "豆包语音 TTS",
"features": "火山引擎豆包语音合成，支持多种音色和情感，国内访问速度快",
"use_case": "需要高质量中文语音合成的用户",
"registration": "https://www.volcengine.com/product/voice-tech"
}
}
@ -147,6 +154,8 @@ def render_tts_settings(tr):
render_qwen3_tts_settings(tr)
elif selected_engine == "indextts2":
render_indextts2_tts_settings(tr)
elif selected_engine == "doubaotts":
render_doubaotts_settings(tr)
# 4. 试听功能
render_voice_preview_new(tr, selected_engine)
@ -703,6 +712,250 @@ def render_indextts2_tts_settings(tr):
config.ui["voice_name"] = f"indextts2:{reference_audio}"
def render_doubaotts_settings(tr):
"""渲染豆包语音 TTS 设置"""
# AK 输入
ak = st.text_input(
"Access Key",
value=config.doubaotts.get("ak", ""),
help="火山引擎 Access Key"
)
# SK 输入
sk = st.text_input(
"Secret Key",
value=config.doubaotts.get("sk", ""),
type="password",
help="火山引擎 Secret Key"
)
# AppID 输入
appid = st.text_input(
"AppID",
value=config.doubaotts.get("appid", ""),
help="豆包语音应用 AppID"
)
# Token 输入
token = st.text_input(
"Token",
value=config.doubaotts.get("token", ""),
type="password",
help="豆包语音应用 Token"
)
# 集群配置
cluster = st.text_input(
"集群",
value=config.doubaotts.get("cluster", "volcano_tts"),
help="业务集群，标准音色使用 volcano_tts"
)
# 音色选择
# 在线音色列表（从文档中提取）
voice_options = {
"BV700_V2_streaming": "灿灿 2.0",
"BV705_streaming": "炀炀",
"BV701_V2_streaming": "擎苍 2.0",
"BV001_V2_streaming": "通用女声 2.0",
"BV700_streaming": "灿灿",
"BV406_V2_streaming": "超自然音色-梓梓2.0",
"BV406_streaming": "超自然音色-梓梓",
"BV407_V2_streaming": "超自然音色-燃燃2.0",
"BV407_streaming": "超自然音色-燃燃",
"BV001_streaming": "通用女声",
"BV002_streaming": "通用男声",
"BV701_streaming": "擎苍",
"BV123_streaming": "阳光青年",
"BV120_streaming": "反卷青年",
"BV119_streaming": "通用赘婿",
"BV115_streaming": "古风少御",
"BV107_streaming": "霸气青叔",
"BV100_streaming": "质朴青年",
"BV104_streaming": "温柔淑女",
"BV004_streaming": "开朗青年",
"BV113_streaming": "甜宠少御",
"BV102_streaming": "儒雅青年",
"BV405_streaming": "甜美小源",
"BV007_streaming": "亲切女声",
"BV009_streaming": "知性女声",
"BV419_streaming": "诚诚",
"BV415_streaming": "童童",
"BV008_streaming": "亲切男声",
"BV408_streaming": "译制片男声",
"BV426_streaming": "懒小羊",
"BV428_streaming": "清新文艺女声",
"BV403_streaming": "鸡汤女声",
"BV158_streaming": "智慧老者",
"BV157_streaming": "慈爱姥姥",
"BR001_streaming": "说唱小哥",
"BV410_streaming": "活力解说男",
"BV411_streaming": "影视解说小帅",
"BV437_streaming": "解说小帅-多情感",
"BV412_streaming": "影视解说小美",
"BV159_streaming": "纨绔青年",
"BV418_streaming": "直播一姐",
"BV142_streaming": "沉稳解说男",
"BV143_streaming": "潇洒青年",
"BV056_streaming": "阳光男声",
"BV005_streaming": "活泼女声",
"BV064_streaming": "小萝莉",
"BV051_streaming": "奶气萌娃",
"BV063_streaming": "动漫海绵",
"BV417_streaming": "动漫海星",
"BV050_streaming": "动漫小新",
"BV061_streaming": "天才童声",
"BV401_streaming": "促销男声",
"BV402_streaming": "促销女声",
"BV006_streaming": "磁性男声",
"BV011_streaming": "新闻女声",
"BV012_streaming": "新闻男声",
"BV034_streaming": "知性姐姐-双语",
"BV033_streaming": "温柔小哥",
"BV511_streaming": "慵懒女声-Ava",
"BV505_streaming": "议论女声-Alicia",
"BV138_streaming": "情感女声-Lawrence",
"BV027_streaming": "美式女声-Amelia",
"BV502_streaming": "讲述女声-Amanda",
"BV503_streaming": "活力女声-Ariana",
"BV504_streaming": "活力男声-Jackson",
"BV421_streaming": "天才少女",
"BV702_streaming": "Stefan",
"BV506_streaming": "天真萌娃-Lily",
"BV040_streaming": "亲切女声-Anna",
"BV516_streaming": "澳洲男声-Henry",
"BV520_streaming": "元气少女",
"BV521_streaming": "萌系少女",
"BV522_streaming": "气质女声",
"BV524_streaming": "日语男声",
"BV531_streaming": "活力男声Carlos（巴西地区）",
"BV530_streaming": "活力女声（巴西地区）",
"BV065_streaming": "气质御姐（墨西哥地区）",
"BV021_streaming": "东北老铁",
"BV020_streaming": "东北丫头",
"BV704_streaming": "方言灿灿",
"BV210_streaming": "西安佟掌柜",
"BV217_streaming": "沪上阿姐",
"BV213_streaming": "广西表哥",
"BV025_streaming": "甜美台妹",
"BV227_streaming": "台普男声",
"BV026_streaming": "港剧男神",
"BV424_streaming": "广东女仔",
"BV212_streaming": "相声演员",
"BV019_streaming": "重庆小伙",
"BV221_streaming": "四川甜妹儿",
"BV423_streaming": "重庆幺妹儿",
"BV214_streaming": "乡村企业家",
"BV226_streaming": "湖南妹坨",
"BV216_streaming": "长沙靓女"
}
saved_voice_type = config.ui.get("doubaotts_voice_type", "BV700_streaming")
if saved_voice_type not in voice_options:
voice_options[saved_voice_type] = f"自定义音色 ({saved_voice_type})"
selected_voice_display = st.selectbox(
"音色选择",
options=list(voice_options.values()),
index=list(voice_options.keys()).index(saved_voice_type) if saved_voice_type in voice_options else 0,
help="选择豆包语音 TTS 音色"
)
# 获取实际的音色ID
voice_type = list(voice_options.keys())[
list(voice_options.values()).index(selected_voice_display)
]
# 高级参数折叠面板
with st.expander("🔧 高级参数", expanded=False):
col1, col2 = st.columns(2)
with col1:
# 语速调节
voice_rate = st.slider(
"语速调节",
min_value=0.2,
max_value=3.0,
value=config.ui.get("doubaotts_rate", 1.0),
step=0.1,
help="调节语音速度 (0.2-3.0)"
)
# 音量调节
voice_volume = st.slider(
"音量调节",
min_value=0.1,
max_value=2.0,
value=config.doubaotts.get("volume", 1.0),
step=0.1,
help="调节语音音量 (0.1-2.0)"
)
with col2:
# 音高调节
voice_pitch = st.slider(
"音高调节",
min_value=0.5,
max_value=1.5,
value=config.doubaotts.get("pitch", 1.0),
step=0.1,
help="调节语音音高 (0.5-1.5)"
)
# 句尾静音时长
silence_duration = st.slider(
"句尾静音时长 (秒)",
min_value=0.0,
max_value=2.0,
value=config.doubaotts.get("silence_duration", 0.125),
step=0.05,
help="调节句尾静音时长 (0.0-2.0秒)"
)
# 显示API Key申请流程
with st.expander("💡 豆包语音 TTS API Key申请流程", expanded=False):
st.write("**申请步骤：**")
st.write("1. 打开 [https://console.volcengine.com/iam/keymanage](https://console.volcengine.com/iam/keymanage)")
st.write("2. 新建 Access Key 和 Secret Key")
st.write("3. 打开 [https://www.volcengine.com/product/voice-tech](https://www.volcengine.com/product/voice-tech)")
st.write("4. 点击立即使用")
st.write("5. 在最左边的API服务中心找到音频生成下面的语音合成（注意：是语音合成，不是语音合成大模型）")
st.write("6. 翻到最下面获取 APPID 和 Access Token")
st.write("")
st.info("💡 请将获取到的 Access Key、Secret Key、AppID 和 Token 填写到上方的配置中")
# 保存配置
config.doubaotts["ak"] = ak
config.doubaotts["sk"] = sk
config.doubaotts["appid"] = appid
config.doubaotts["token"] = token
config.doubaotts["cluster"] = cluster
config.doubaotts["volume"] = voice_volume
config.doubaotts["pitch"] = voice_pitch
config.doubaotts["silence_duration"] = silence_duration
config.ui["doubaotts_voice_type"] = voice_type
config.ui["doubaotts_rate"] = voice_rate
config.ui["voice_name"] = voice_type # 兼容性
st.session_state['voice_rate'] = voice_rate # 确保语速参数被保存到session state
# 显示配置状态
if ak and sk and appid and token:
st.success("✅ 豆包语音 TTS 配置已设置")
else:
missing = []
if not ak:
missing.append("Access Key")
if not sk:
missing.append("Secret Key")
if not appid:
missing.append("AppID")
if not token:
missing.append("Token")
if missing:
st.warning(f"⚠️ 请配置: {', '.join(missing)}")
def render_voice_preview_new(tr, selected_engine):
"""渲染新的语音试听功能"""
if st.button("🎵 试听语音合成", use_container_width=True):
@ -746,6 +999,11 @@ def render_voice_preview_new(tr, selected_engine):
voice_name = f"indextts2:{reference_audio}"
voice_rate = 1.0 # IndexTTS2 不支持速度调节
voice_pitch = 1.0 # IndexTTS2 不支持音调调节
elif selected_engine == "doubaotts":
voice_type = config.ui.get("doubaotts_voice_type", "BV700_streaming")
voice_name = voice_type
voice_rate = config.ui.get("doubaotts_rate", 1.0)
voice_pitch = 1.0 # 豆包语音 TTS 不支持音调调节
if not voice_name:
st.error("请先配置语音设置")