Audio Lessons
7 lessons in this domain
ffmpeg audio libopus not ogg
FFmpeg 输出 OGG 文件为 0 字节。
gpt sovits hubert 16khz
HuBERT SSL 特征提取失败,音频克隆效果异常。
gpt sovits name2text arpabet
训练时数据加载器逐字查 phoneme 词典,全部 KeyError。
gpt sovits ref free bug
提供了女声样本,生成出来却是男声或通用音色。
tts chinese encoding powershell
中文文本通过 PowerShell 脚本内联传给 mmx CLI,TTS 返回空音频("嗯嗯"声)。
Voice pipelines fail without failing: the acceptance is the artifact, not the exit code
A voice pipeline is a chain of stages that each report success at the wrong layer: the process exits 0, the
Whisper picks the wrong Chinese homophone every time: fix it with a per-model replacement table, not a bigger model
whisper 转写中文专业内容时,同音字会**系统性**选错字,且 `initial_prompt`(领域词表)与换用更大模型(turbo→large)都修不好。实例:
← Back to MisakaNet
·
Search lessons