https://github.com/ideasman42/nerd-dictation
Simple, hackable offline speech to text - using the VOSK-API.
https://github.com/ideasman42/nerd-dictation
Last synced: 2 days ago
JSON representation
Simple, hackable offline speech to text - using the VOSK-API.
- Host: GitHub
- URL: https://github.com/ideasman42/nerd-dictation
- Owner: ideasman42
- License: gpl-3.0
- Created: 2021-05-25T16:04:37.000Z (almost 4 years ago)
- Default Branch: main
- Last Pushed: 2025-04-01T05:44:07.000Z (14 days ago)
- Last Synced: 2025-04-05T19:04:36.563Z (9 days ago)
- Language: Python
- Homepage:
- Size: 128 KB
- Stars: 1,408
- Watchers: 27
- Forks: 114
- Open Issues: 64
-
Metadata Files:
- Readme: readme-sox.rst
- Changelog: changelog.rst
- License: LICENSE
Awesome Lists containing this project
- stars - ideasman42/nerd-dictation - Simple, hackable offline speech to text - using the VOSK-API. (Python)
README
#################################
Using ``sox`` with nerd-dictation
#################################
This guide explains how to configure ``sox`` for recording audio with ``nerd-dictation``.
When should I use ``sox``?
==========================
You may wish to configure ``sox`` if you are using a system without pulse-audio support (such as FreeBSD).
Configuring ``sox``
===================
Set environment variable ``AUDIODEV`` to use a specific input device.
Other sox options can be set (such as gain) by setting environment variable ``SOX_OPTS``.
You can test various devices by
.. code-block:: sh
# List audio devices.
arecord -l || cat /proc/asound/cards || cat /dev/sndstat
# Example, use card 2, subdevice 0.
# Record 10 seconds and playback to default output.
AUDIODEV='hw:2,0' sox -d --buffer 1000 -r 16000 -b 16 -e signed-integer -c 1 -t wav -L test.wav trim 0 10
sox test.wav -d
Running nerd dictation with ``sox``
===================================
To run ``nerd-dictation`` with ``sox``, use the ``--input`` argument.
.. code-block:: sh
nerd-dictation begin --input=SOX