AnkiWeb addon 903988322
Generates high-quality Anki audio via a locally hosted Kokoro text-to-speech model or other OpenAI-compatible TTS services, supporting nine languages.
Open on AnkiWeb GitHub Find alternatives
stale
| Min Anki | Max Anki | Updated |
|---|---|---|
| 2.1.50 | 24.04+ | 2025-02-13 |
Loadingโฆ
Generates high-quality Japanese text-to-speech audio for Anki cards through VOICEVOX, letting users right-click cards in the browser and choose source/destination fields, speaker, and style.
Adds speech to Anki flashcards using numerous free and premium text-to-speech services, displaying all voices together for easy preview, comparison, and audio generation.
Adds automated translation for 133 languages, transliteration for 33, and text-to-speech with 547 voices across 63 languages to Anki, requiring a paid API key.
Generates AI-assisted text and Edge TTS audio from Anki note fields using configurable OpenAI-compatible providers, with prompt templates, batch processing, and optional image OCR input.
Adds high-quality neural text-to-speech audio to individual cards or entire decks in 53 languages, with both single-card and whole-deck generation options.
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.
Kokoro TTS is the latest lightweight and high-performance text-to-speech (TTS) model for 2025. The generated audio is almost indistinguishable from that of a real person in terms of tone and intonation. You can give it a try at this link.
What's even more impressive is that its model parameters are minimal, which means it can run smoothly on a regular CPU and GPU in a home computer. Moreover, it supports generating American English, British English, Spanish, French, Hindi, Italian, Japanese, Brazilian Portuguese, and Mandarin Chinese.
I've been using the AwesomeTTS plugin to generate audio in Anki with the free service, and I've always felt that the quality of the generated audio wasn't up to par. So, I installed the Kokoro TTS model locally, and also made improvements to the AwesomeTTS, adding the ability to call the local Kokoro service.
First, you need to disable or remove the AwesomeTTS plugin from the Anki plugin page. Then, enter the plugin code 903988322 to install the AwesomeTTS-AI-Version (unofficial).
Now, in order to download KokoroTTS model, if you are a programmer and familiar with Docker, please refer to this tutorial. Otherwise, please see this tutorial.
Kokoro TTS is the latest lightweight and high-performance text-to-speech (TTS) model for 2025. The generated audio is almost indistinguishable from that of a real person in terms of tone and intonation. You can give it a try at this link. What's even more impressive is that its model parameters are minimal, which means it can run smoothly on a regular CPU and GPU in a home computer. Moreover, it supports generating American English, British English, Spanish, French, Hindi, Italian, Japanese, Brazilian Portuguese, and Mandarin Chinese.
I've been using the AwesomeTTS plugin to generate audio in Anki with the free service, and I've always felt that the quality of the generated audio wasn't up to par. So, I installed the Kokoro TTS model locally, and also made improvements to the AwesomeTTS, adding the ability to call the local Kokoro service.
Since the author of AwesomeTTS has stopped updating AwesomeTTS and has shifted to maintaining another project called HyperTTS, you can completely disable or remove the AwesomeTTS plugin from the Anki plugin page without any worries. Then, enter the plugin code 903988322 to install the AwesomeTTS-AI-Version (unofficial).
After installing the plugin, you also need to download Kokoro's model using Docker locally. Enter one of the following two commands in the command line, then configure the plugin to access the Kokoro API address, which is default to http://localhost:8888/v1/audio/speech. Note that if you change the container's port, make sure to update the port number in the API address.
docker run -p 8888:8880 ghcr.io/remsky/kokoro-fastapi-cpu:v0.2.0 # CPU, or:
docker run --gpus all -p 8888:8880 ghcr.io/remsky/kokoro-fastapi-gpu:v0.2.0 #NVIDIA GPU
Here is a detailed tutorial for everyone
Additionally, I've added OpenAI Compatible (OpenAI Interface Compatible) service, which means you can install other OpenAI compatible TTS models locally (such as openai-edge-tts) for tts service.
Be mindful that when using OpenAI Compatible, there may be instances where you've selected OpenAI Compatible, but the generated voice turns out to be Kokoro's. This is likely due to a caching issue; providing it with a new word to generate should refresh the cache.
AwesomeTTS-AI-Version-unofficial-: GPL-3.0 license anki-awesome-tts: GPL-3.0 license