1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
|
Usage: dmvop [OPTIONS]
Options
--device=<name> Microphone (WASAPI name or device ID). Required.
-O, --output=<mode> Output destination(s). May be repeated. [default: stdout]
stdout | stderr | tcp | udp | udp-broadcast | ipc
-m, --model=<model> Whisper model. [default: base_en]
--lang=<code> Language hint (zh, ja, en, fr...). Skips detection.
--post=<spec> Post-process transcribed text, e.g. +pinyin
--models-dir=<path> Directory for model files. Fails if invalid.
-f, --format=<pattern> Output format with %{vol}, %{word}, %{confid}.
[default: %{vol},%{word}]
-S, --format-file=<path> Read format from file.
-p, --port=<num> Port for TCP/UDP modes. [default: 5117]
--socket-file=<path> Unix socket path for IPC. [default: ./dmvop.sock]
--subnet-mask=<mask> Subnet mask for UDP broadcast.
[default: 255.255.255.0]
-L, --list-devices List microphones and exit.
--list-models List Whisper models and exit.
--config=<path> Config file (TOML). Falls back to ./dmvop.toml.
--download-model=<name> Download a model and exit.
--instant Aggressive mode — shorter segments for
near-real-time output. Speak slowly.
-V, --verbose Show debug output.
-h, --help Show this help.
Examples
dmvop --list-devices
dmvop --device="My Mic" -O stdout -O tcp -m small --lang=zh
dmvop --device="My Mic" -f "[%{vol}] %{word}"
dmvop --device="My Mic" -O udp-broadcast -p 8888
dmvop --download-model=small
Notes
- First run downloads the model (~142 MB for base_en).
- Use --verbose to see speech events and timing.
- Ctrl+C to stop.
- All outputs initialize in parallel.
|
