Sentence Audio Generator (AI TTS)

AnkiWeb addon 510081046

Generates sentence audio via a local AI TTS server and bulk-translates text fields, processing only notes selected in Browse and writing WAV files into mapped fields.
AI-generated summary; may contain mistakes.

card-creationainote-managementbeta

Open on AnkiWeb GitHub Find alternatives

AnkiWeb

Votes
0 (👍 0 · 👎 0)
Updated
2026-04-23
Anki versions
25.09~
License
Not found (this does not mean that the addon has no license)
Virus scan
No ClamAV match A match does not prove malware. No match does not prove that the code is safe.
Languages
Python, JSON, CSV
Description language
en

Maintenance

stale

  • Last update or commit: 2026-04-27 (188 days before the snapshot).
  • Commits in the last year: 4.
  • Issues: 0 open, 0 closed.
  • Pull requests: 0 open, 0 merged.

Will it work on my Anki?

Version branches
Min AnkiMax AnkiUpdated
25.09+2026-04-23
History across monthly snapshots

Loading…

Similar addons

AI Language Explainer

Generates contextual written explanations and spoken audio for target words in sentence cards using OpenAI models and selectable TTS engines, with one-click or batch processing.

card-creationaisync-integrationdictionary

10 votes 72% approval 👍 10 · 👎 0 ⭐ 9 Anki 2.1.66~ Updated 2025-10-07
Japanese Support - Auto Fill cards

Automatically fills Japanese vocabulary and sentence flashcards from a typed word or sentence, adding translations, kana readings, kanji readings, transitivity, part-of-speech, offline dictionary lookups, and Google TTS audio.

card-creationcard-templatesdictionary

1 votes 21% approval 👍 1 · 👎 0 Anki 2.1.1~ Updated 2026-07-28

Description

Generate sentence audio using AI TTS.

Requirements:

  • Nvidia GPU with 8GB of VRAM or more (for the chosen TTS Engine)

Designed to be used with Japanese to generate audio for a given sentence. The configuration needs to be set up for at least 1 deck type and note type, where the input field is a sentence and the output field is an audio field.

The intended TTS engine is Irodori-TTS, the best quality-for-speed engine that is open source. It has very strong voice cloning capabilities, allowing you to use any character's voice to read out your sentence. It works best if the input audio file is in Japanese.

To utilize the add-on, Irodori needs to be installed (https://github.com/JarodMica/Irodori-TTS-API) and the server running in order for it to generate audio.

More details later, WIP.

README

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.