vocast 0.1.0

Convert articles to audio you can stream, using local TTS models

Vocast

Convert articles to audio using local TTS models.

Why

I wanted a way to convert articles into audio that I could stream to my mobile device while on the go. This seemed straightforward enough to build myself and I didn't want to pay for an app.

How

Vocast uses Kokoro for TTS. It can fetch articles from a given URL or local text file. Audio files are saved to ~/.vocast/library. It provides an HTTP server that exposes an RSS feed allowing for podcast apps to discover the converted audio files. You can use Tailscale to allow connections between the server and client devices like your mobile phone. You will need to use a podcast app that does not proxy requests through their own servers as that will prevent the app from connecting to your Vocast server.

Requirements

• Python 3.10–3.12 (Kokoro does not yet support 3.13). Installing with uv provisions a compatible Python for you.
• espeak-ng on PATH (used by Kokoro as a fallback phonemizer)

ffmpeg is bundled (via imageio-ffmpeg), so the only system dependency is espeak-ng:

sudo dnf install espeak-ng     # Fedora
sudo apt install espeak-ng     # Debian/Ubuntu
brew install espeak-ng         # macOS

Install

The recommended way is an isolated tool install, which also provisions a compatible Python for you and puts a vocast command on your PATH:

uv tool install git+https://github.com/cnrmurphy/vocast
# or, with pipx:
pipx install git+https://github.com/cnrmurphy/vocast

(Once published to PyPI this becomes simply uv tool install vocast.)

The first run downloads the Kokoro weights (~300 MB) and a small spaCy model into the cache. Subsequent runs are immediate.

From source (development)

uv venv && source .venv/bin/activate
uv pip install -e .

Usage

Vocast has subcommands; run vocast --help to see them all.

Add an article to the library

vocast add https://example.com/article    # fetch and synthesize a URL
vocast add notes.txt                      # synthesize a local text file
vocast add ... --title "Custom title"     # override the title
vocast add ... --voice af_bella           # use a different Kokoro voice
vocast add ... --quiet                    # suppress per-chunk progress

URLs are fetched and cleaned with trafilatura. Code blocks (<pre> elements) are stripped before synthesis since they don't translate well to audio. Each entry is stored under ~/.vocast/library/<id>/ as audio.mp3 plus meta.json with title, source URL, duration, and voice.

List the library

vocast list

Serve the library as a podcast feed

vocast serve                                # 127.0.0.1:8080 by default
vocast serve --host 0.0.0.0 --port 8000     # custom host/port

Exposes GET /feed.xml (podcast RSS) and GET /audio/<id>.mp3 (audio enclosures). Long articles are split on sentence boundaries during synthesis and concatenated with short silence between chunks.

Expose the feed to your phone over Tailscale

vocast init

A guided checklist that walks you through installing Tailscale, signing into your tailnet, and proxying vocast serve over HTTPS via tailscale serve. Re-run after each step until it prints your feed URL, then add that URL to a direct-download podcast app (Downcast is confirmed working).

Synthesize directly to a file (skip the library)

vocast synth article.txt              # writes article.mp3
vocast synth article.txt -o out.wav   # WAV output