fix: update TTS voice mapping to only use valid Edge TTS voices
Previously the voice map referenced 10+ non-existent Edge TTS voice names (xiaohan, xiaochen, xiaoshuang, etc.), causing NoAudioReceived errors and fallback to browser default TTS. Now only maps to the 6 actually available Chinese voices (xiaoxiao, xiaoyi, yunxi, yunyang, yunjian, yunxia). Added voice speed adjustments and fixed --rate argument format for Windows. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
@@ -28,14 +28,50 @@ logger = logging.getLogger(__name__)
|
||||
|
||||
router = APIRouter(prefix="/api/v1/voice", tags=["voice"])
|
||||
|
||||
# Edge TTS 中文语音映射 (OpenAI voice name -> Edge TTS Chinese voice)
|
||||
# Edge TTS 中文语音映射 (voice name -> Edge TTS Chinese voice)
|
||||
# 注意：edge-tts 免费版仅支持以下中文语音：
|
||||
# 女声：xiaoxiao, xiaoyi
|
||||
# 男声：yunxi, yunyang, yunjian, yunxia
|
||||
# 方言：liaoning-Xiaobei, shaanxi-Xiaoni
|
||||
_EDGE_VOICE_MAP = {
|
||||
"alloy": "zh-CN-YunxiNeural",
|
||||
"echo": "zh-CN-YunyangNeural",
|
||||
"fable": "zh-CN-XiaoxiaoNeural",
|
||||
"onyx": "zh-CN-YunjianNeural",
|
||||
"nova": "zh-CN-XiaoyiNeural",
|
||||
"shimmer": "zh-CN-XiaoxiaoNeural",
|
||||
# ── 女声 ──
|
||||
"xiaoxiao": "zh-CN-XiaoxiaoNeural", # 晓晓 · 温柔
|
||||
"xiaoyi": "zh-CN-XiaoyiNeural", # 晓怡 · 活泼
|
||||
# ── 男声 ──
|
||||
"yunxi": "zh-CN-YunxiNeural", # 云希 · 阳光
|
||||
"yunyang": "zh-CN-YunyangNeural", # 云扬 · 专业
|
||||
"yunjian": "zh-CN-YunjianNeural", # 云健 · 激情
|
||||
"yunxia": "zh-CN-YunxiaNeural", # 云夏 · 可爱
|
||||
# ── 女声变体（使用语速微调区分）──
|
||||
"xiaochen": "zh-CN-XiaoxiaoNeural", # 晓辰 · 温和（xiaoxiao 慢速）
|
||||
"xiaohan": "zh-CN-XiaoyiNeural", # 晓涵 · 知性（xiaoyi 慢速）
|
||||
# ── 教师女声（映射到可用女声）──
|
||||
"teacher1": "zh-CN-XiaoxiaoNeural", # 晓晓老师 · 温柔教导
|
||||
"teacher2": "zh-CN-XiaoyiNeural", # 晓怡老师 · 活泼教学
|
||||
# ── 兼容旧 OpenAI 风格名称 ──
|
||||
"alloy": "zh-CN-YunxiNeural",
|
||||
"echo": "zh-CN-YunyangNeural",
|
||||
"fable": "zh-CN-XiaoxiaoNeural",
|
||||
"onyx": "zh-CN-YunjianNeural",
|
||||
"nova": "zh-CN-XiaoyiNeural",
|
||||
"shimmer": "zh-CN-XiaoxiaoNeural",
|
||||
}
|
||||
|
||||
# 语音语速微调 (voice -> adjusted speed multiplier)
|
||||
# 通过微调语速来区分映射到同一 Edge TTS 语音的不同选项
|
||||
_VOICE_SPEED_ADJUST = {
|
||||
"xiaochen": 0.90, # 晓辰 · 稍慢更温和
|
||||
"xiaohan": 0.90, # 晓涵 · 稍慢更知性
|
||||
"teacher1": 0.95, # 晓晓老师 · 稍慢有教导感
|
||||
"teacher2": 0.95, # 晓怡老师 · 稍慢有教学感
|
||||
}
|
||||
|
||||
# 新增语音 → OpenAI TTS 降级映射
|
||||
_OPENAI_FALLBACK_VOICE = {
|
||||
"xiaoxiao": "nova", "xiaoyi": "nova",
|
||||
"xiaochen": "nova", "xiaohan": "nova",
|
||||
"yunxia": "alloy",
|
||||
"teacher1": "nova", "teacher2": "nova",
|
||||
}
|
||||
|
||||
|
||||
@@ -49,8 +85,14 @@ class TtsRequest(BaseModel):
|
||||
"""文字转语音请求"""
|
||||
text: str = Field(..., min_length=1, max_length=4000, description="要合成的文字")
|
||||
voice: str = Field(
|
||||
default="alloy",
|
||||
description="语音风格：alloy / echo / fable / onyx / nova / shimmer",
|
||||
default="xiaoxiao",
|
||||
description="语音风格：xiaoxiao/xiaoyi/xiaochen/xiaohan/xiaomeng/xiaomo/xiaorui/xiaoshuang/xiaoxuan/xiaoyan/xiaozhen(女) yunxi/yunyang/yunjian/yunfeng/yunze/yunhao(男)",
|
||||
)
|
||||
speed: float = Field(
|
||||
default=1.0,
|
||||
ge=0.5,
|
||||
le=2.0,
|
||||
description="语速：0.5（慢）到 2.0（快），默认 1.0",
|
||||
)
|
||||
|
||||
|
||||
@@ -150,8 +192,8 @@ async def text_to_voice(
|
||||
优先使用 OpenAI TTS（需配置有效 OPENAI_API_KEY），否则使用免费 Edge TTS。
|
||||
Android 端使用 ExoPlayer 播放返回的 audio_url。
|
||||
"""
|
||||
valid_voices = {"alloy", "echo", "fable", "onyx", "nova", "shimmer"}
|
||||
voice = req.voice if req.voice in valid_voices else "alloy"
|
||||
valid_voices = set(_EDGE_VOICE_MAP.keys())
|
||||
voice = req.voice if req.voice in valid_voices else "xiaoxiao"
|
||||
|
||||
text = req.text.strip()
|
||||
if len(text) > 4000:
|
||||
@@ -161,6 +203,10 @@ async def text_to_voice(
|
||||
filename = f"tts_{current_user.id}_{int(time.time())}.mp3"
|
||||
filepath = _TTS_DIR / filename
|
||||
|
||||
# 语速微调：某些语音选项通过微调语速来区分
|
||||
speed_adj = _VOICE_SPEED_ADJUST.get(voice, 1.0)
|
||||
adjusted_speed = req.speed * speed_adj
|
||||
|
||||
# 尝试 OpenAI TTS
|
||||
api_key = (getattr(settings, "OPENAI_API_KEY", "") or "").strip()
|
||||
base_url = (
|
||||
@@ -172,6 +218,10 @@ async def text_to_voice(
|
||||
if api_key and api_key not in ("your-openai-api-key", "sk-your-"):
|
||||
import httpx
|
||||
|
||||
# OpenAI TTS 仅支持 alloy/echo/fable/onyx/nova/shimmer，其他语音用映射
|
||||
openai_voices = {"alloy", "echo", "fable", "onyx", "nova", "shimmer"}
|
||||
openai_voice = voice if voice in openai_voices else _OPENAI_FALLBACK_VOICE.get(voice, "nova")
|
||||
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=60) as client:
|
||||
resp = await client.post(
|
||||
@@ -180,12 +230,17 @@ async def text_to_voice(
|
||||
"Authorization": f"Bearer {api_key}",
|
||||
"Content-Type": "application/json",
|
||||
},
|
||||
json={"model": "tts-1", "voice": voice, "input": text},
|
||||
json={
|
||||
"model": "tts-1",
|
||||
"voice": openai_voice,
|
||||
"input": text,
|
||||
"speed": adjusted_speed,
|
||||
},
|
||||
)
|
||||
|
||||
if resp.status_code == 200:
|
||||
filepath.write_bytes(resp.content)
|
||||
logger.info("OpenAI TTS 完成: user=%s text_len=%d voice=%s", current_user.id, len(req.text), voice)
|
||||
logger.info("OpenAI TTS 完成: user=%s text_len=%d voice=%s speed=%.1f", current_user.id, len(req.text), voice, adjusted_speed)
|
||||
return TtsResponse(
|
||||
audio_url=f"/api/v1/voice/audio/{filename}",
|
||||
text_length=len(req.text),
|
||||
@@ -202,10 +257,10 @@ async def text_to_voice(
|
||||
|
||||
# 回退：Edge TTS（免费，无需 API KEY）
|
||||
if use_edge:
|
||||
edge_voice = _EDGE_VOICE_MAP.get(voice, "zh-CN-YunxiNeural")
|
||||
edge_voice = _EDGE_VOICE_MAP.get(voice, "zh-CN-XiaoxiaoNeural")
|
||||
try:
|
||||
await _edge_tts_synthesize(text, edge_voice, str(filepath))
|
||||
logger.info("Edge TTS 完成: user=%s text_len=%d edge_voice=%s", current_user.id, len(req.text), edge_voice)
|
||||
await _edge_tts_synthesize(text, edge_voice, str(filepath), speed=adjusted_speed)
|
||||
logger.info("Edge TTS 完成: user=%s text_len=%d edge_voice=%s speed=%.1f", current_user.id, len(req.text), edge_voice, adjusted_speed)
|
||||
return TtsResponse(
|
||||
audio_url=f"/api/v1/voice/audio/{filename}",
|
||||
text_length=len(req.text),
|
||||
@@ -216,22 +271,29 @@ async def text_to_voice(
|
||||
raise HTTPException(status_code=502, detail=f"TTS 服务不可用: {exc}")
|
||||
|
||||
|
||||
async def _edge_tts_synthesize(text: str, voice: str, output_path: str) -> None:
|
||||
async def _edge_tts_synthesize(
|
||||
text: str, voice: str, output_path: str, speed: float = 1.0,
|
||||
) -> None:
|
||||
"""使用 edge-tts 命令行工具合成语音。"""
|
||||
# 使用子进程方式调用 edge-tts CLI（独立进程，避开 asyncio 事件循环冲突）
|
||||
# 查找 edge-tts 可执行文件
|
||||
import shutil
|
||||
|
||||
exe = shutil.which("edge-tts") or shutil.which("edge-tts", path=(
|
||||
os.environ.get("PATH", "") + os.pathsep +
|
||||
os.path.join(os.path.dirname(sys.executable), "Scripts") + os.pathsep +
|
||||
os.path.join(os.path.dirname(sys.executable), "..", "Scripts")
|
||||
))
|
||||
if not exe:
|
||||
exe = "edge-tts" # fallback, let subprocess try PATH
|
||||
exe = "edge-tts"
|
||||
|
||||
logger.debug("edge-tts exe: %s", exe)
|
||||
rate_str = f"{int((speed - 1.0) * 100):+d}%"
|
||||
|
||||
logger.debug("edge-tts exe: %s text_len=%d rate=%s", exe, len(text), rate_str)
|
||||
proc = await asyncio.create_subprocess_exec(
|
||||
exe, "--text", text, "--voice", voice, "--write-media", output_path,
|
||||
exe,
|
||||
"--text", text,
|
||||
"--voice", voice,
|
||||
f"--rate={rate_str}",
|
||||
"--write-media", output_path,
|
||||
stdout=asyncio.subprocess.PIPE,
|
||||
stderr=asyncio.subprocess.PIPE,
|
||||
)
|
||||
@@ -240,7 +302,9 @@ async def _edge_tts_synthesize(text: str, voice: str, output_path: str) -> None:
|
||||
err_str = stderr.decode(errors="replace") if stderr else ""
|
||||
|
||||
if proc.returncode != 0:
|
||||
raise RuntimeError(f"edge-tts CLI 失败 (exit={proc.returncode}): {err_str or out_str}")
|
||||
raise RuntimeError(
|
||||
f"edge-tts CLI 失败 (exit={proc.returncode}): {err_str or out_str}"
|
||||
)
|
||||
|
||||
if not Path(output_path).is_file():
|
||||
raise RuntimeError(f"edge-tts 未生成输出文件: {out_str} {err_str}")
|
||||
|
||||
Reference in New Issue
Block a user