Yomichan Audio Server

Although Yomichan does support custom audio URLs, it doesn't solve the problem of using a local directory. Instead of having to mess with nginx configs, with yomi-audio, you can easily start an audio server and pass it to Yomichan to pull pronunciation directly from your local filesystem.

Setup

Python and pip are required to run the server. Clone the repository and install the necessary dependencies:

git clone https://github.com/kamui-fin/yomi-audio && cd yomi-audio
pip install -r requirements.txt

Running locally

The script offers a CLI that customizes the behavior of the server and starts it. The main required option is -d or --directories to specify the location of one or more directories containing the audio files. For example:

python app.py -d /home/user/audio_bank

Sorting

By default, the original order in which they were specified is retained without any sorting.

With the -s flag, the directories passed in are sorted in ascending order by name. This can be leveraged to prioritize results from directories that contain higher quality audio.

Pattern matching

The default pattern for matching audio files is {term}.mp3. Obviously, this may not suffice depending on the naming scheme. As a result, there are two different ways to customize the way the correct file is found.

With both ways, the term and reading must be surrounded by {} as a template literal, where they will be later replaced during matching.

One way is to use -p to directly match using a completely static/pre-defined naming pattern. This can be useful if the underlying naming scheme of the directory does not vary. For example, if all the files are named like pronunciation_喜欢.mp3, you can simply use the static pattern pronunciation_{term}.mp3.

The other way with -g utilizes powerful glob functionality. An example would be ???{term}???.mp3 to match {term} with 3 leading and trailing characters.

Connecting to Yomichan

Once you have the server running, open Yomichan settings. Click "Configure audio playback sources" under the Audio section and add the following custom URL:

http://localhost:5000/?expression={term}&reading={reading}