Full Parameter Reference

Global Options

Option	Description	Default	Env Var
`-b, --base <path>`	Model root dir	Program Dir	`MANYSPEECH_BASE`
`-o, --output <dir>`	Working dir	Current Dir	`MANYSPEECH_OUTPUT`
`--accuracy <fp32/int8>`	Model precision	`int8`	`MANYSPEECH_ACCURACY`
`--threads <num>`	Inference threads, 0=Auto	`2`	`MANYSPEECH_THREADS`
`--debug`	Enable debug output	—	—
`--save-default`	Save params as default	—	—
`--show-config`	Show current config	—	—
`--reset-config`	Reset config to defaults	—	—

Option	Description	Default	Env Var
`-t, --type <online/offline/2pass>`	Recognition type (Required)	—	`MANYSPEECH_TYPE`
`-m, --method <one/batch/chunk>`	Processing method	`chunk`	`MANYSPEECH_METHOD`
`-i, --input <file/mic>`	Input source	`file`	`MANYSPEECH_INPUT`
`--format <json/text/srt/vtt>`	Output format	`text`	`MANYSPEECH_FORMAT`
`-f, --files <files...>`	Audio files (Multi-support)	—	—
`-md, --model <name>`	Main model name	Auto	`MANYSPEECH_MODEL`
`--model2 <name>`	2pass secondary model	Auto	`MANYSPEECH_MODEL2`
`--vad <name>`	VAD model name	`alifsmnvad-onnx`	`MANYSPEECH_VAD`
`--punc <name>`	Punctuation model	`alicttransformerpunc-zh-en-mge-int8-onnx`	`MANYSPEECH_PUNC`

Option	Description	Default
`--text <text>`	Text to punctuate	—
`-f, --files <files...>`	Input text files	—
`--split <num>`	Text split size	`15`
`--format <json/text>`	Output format	`text`
`--punc <name>`	Punctuation model	`alicttransformerpunc-zh-en-mge-int8-onnx`