Sentence Audio Generator (AI TTS)

Sentence Audio Generator (AI TTS) is an Anki add-on that processes selected notes in the Browse window. It can generate audio by sending sentence text to a configurable local HTTP TTS service and writing the returned WAV audio into a target field as an Anki [sound:...] tag. It can also translate source-field text into a target text field through a local vLLM-compatible completions endpoint.

What It Does

• Adds Bulk add TTS to the Browse window Edit menu.
• Adds Bulk translate sentences to the Browse window Edit menu.
• Processes only the notes currently selected in Browse.
• Uses deck + note type mappings to choose the source text field and target audio field.
• Sends requests to a configurable local TTS endpoint.
• Sends translation requests to a configurable local vLLM completions endpoint.
• Stores a global reference audio file and optional transcript file under the add-on's user_files/ folder.
• Skips notes with empty source text.
• Skips notes with existing target audio unless overwrite is enabled for existing TTS-tagged audio.
• Skips notes with existing translation target text unless translation overwrite is enabled.

Requirements

• Anki 25.09.2 or a compatible recent Anki 25 build.
• Windows-focused local workflow.
• A running local TTS server that accepts multipart form uploads and returns audio/wav bytes.
• Optional: a running local vLLM-compatible translation server that accepts JSON completions requests.

Build

Use the repo-local build script to package the add-on:

build.bat

build.bat:

• reads app_version.txt
• optionally updates the version when prompted
• requires the version to use X.Y.Z semantic form
• rebuilds dist\SentenceAudioGenerator.ankiaddon
• rebuilds dist\SentenceAudioGenerator-<version>.ankiaddon

Use this generated package when installing or updating the add-on:

dist\SentenceAudioGenerator.ankiaddon

Install And Setup In Anki

1. Open Tools -> Add-ons.
2. Install or update the add-on from AnkiWeb.
3. Select the add-on and click Config.
4. Set the endpoint values.
5. Import the global reference audio file.
6. Optionally import a transcript text file.
7. Add at least one deck + note type mapping for TTS.
8. Optionally configure the Translation tab and add translation mappings.
9. Save the settings dialog.

The add-on registers a custom config dialog through Anki's add-on manager, so the Config button should open the Qt settings dialog instead of the raw JSON editor.

After configuration, open Browse, select a small set of notes, and run Edit -> Bulk add TTS or Edit -> Bulk translate sentences.

Default Irodori Endpoint Settings

These defaults match the current local Irodori profile used in development:

• Host: 127.0.0.1
• Port: 9150
• Path: /tts
• Text form field: text
• Reference audio form field: reference_audio
• Transcript text form field: blank by default

Default Translation Endpoint Settings

These defaults match the current local TranslateGemma/vLLM profile used in development:

• Host: 127.0.0.1
• Port: 8009
• Path: /v1/completions
• Model: translategemma-4b-it
• Max output tokens: 256

Generated Media

Generated files are written into Anki media with names shaped like:

tts-<note_id>-<sha1_prefix>.wav

The tts prefix is also used by the overwrite option to distinguish generated TTS audio from other audio in the target field.