README file from
GithubThe README couldn't be loaded right now. View it on GitHub →
Description
Read notes aloud with system voices and export them as WAV with a local, downloadable German voice. No cloud, no account.
Reviews
No reviews yet.
Stats
0
stars
66
downloads
0
forks
29
days
18
days
29
days
0
total PRs
0
open PRs
0
closed PRs
0
merged PRs
0
total issues
0
open issues
0
closed issues
56
commits
Latest Version
a month ago
Changelog
Added
- A second downloadable voice: English (Piper LJSpeech, en_US, medium, dataset public domain) next to the German Thorsten. A new setting Voice for export switches between them; it also applies to reading aloud with the downloaded voice.
- On first start the voice follows Obsidian's display language (English UI → English voice). A choice you have made is never overwritten.
Changed
- Worker and ONNX runtime are now shared between voices: a second voice downloads only its model (≈ 60 MB instead of ≈ 75 MB), the voice list shows what each one still costs, and Remove keeps the shared runtime as long as another voice needs it.
- The synthesis worker carries both eSpeak-NG dictionaries (English and German), ≈ 2.8 MB instead of ≈ 2.2 MB; only the one the selected voice needs is loaded at runtime.