| domain | contrib |
|---|---|
| title | gpt sovits name2text arpabet |
| verification | metadata-normalized |
| {"title" | GPT-SoVITS 训练:2-name2text 格式必须用 ARPABET 音素而非中文原文", "domain": "tts", "tags": "", "source": "hanged-man", "status": "published", "created": "2026-04-06", "confidence": "0.9", "scope": "narrow", "domain_expert": "hanged-man", "verified_date": "2026-04-06"} |
| created | 2026-07-06 |
| source | unknown |
训练时数据加载器逐字查 phoneme 词典,全部 KeyError。
2-name2text.txt 第二列误写为中文原文,正确应为 ARPABET 音素符号(空格分隔)。
basename {w o2 h en3 AA ai4 ...} {type} {language}
注意:
- 中文→ARPABET:用
g2p(text_normalize(text)) - 必须先
text_normalize再 g2p,中文标点(,。)需先规范化为 ASCII 标点 - 音频文件必须加
.wav扩展名,无扩展名的 WAV ffmpeg 无法识别
- Follow the solution steps in order
- Run any relevant commands or tests to confirm the fix
- Verify the symptom no longer occurs
- Check related logs or outputs for expected behavior
音素训练数据格式必须严格按文档,词典只认音标不认文字。