https://github.com/yanorei32/aitalked-server

Simple VOICEROID / GynoidTalk Web Server based on aitalked library
https://github.com/yanorei32/aitalked-server

aitalked api axum gynoidtalk japanese japanese-tts rust text-to-speech tts voiceroid voiceroid2 web

Last synced: about 1 month ago
JSON representation

Simple VOICEROID / GynoidTalk Web Server based on aitalked library

• Host: GitHub
• URL: https://github.com/yanorei32/aitalked-server
• Owner: yanorei32
• License: bsd-2-clause
• Created: 2025-09-26T03:22:02.000Z (9 months ago)
• Default Branch: master
• Last Pushed: 2026-05-20T01:42:04.000Z (about 1 month ago)
• Last Synced: 2026-05-20T04:43:09.415Z (about 1 month ago)
• Topics: aitalked, api, axum, gynoidtalk, japanese, japanese-tts, rust, text-to-speech, tts, voiceroid, voiceroid2, web
• Language: Rust
• Homepage:
• Size: 247 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# aitalked server

This is a lightweight web server designed for internal use with [Discord TTS](https://github.com/yanorei32/discord-tts), enabling headless synthesis of speech using VOICEROID and Gynoidtalk engines.



## Features

- 🗣️ **Headless Speech Synthesis**  

  Utilizes the [aitalked](https://github.com/yanorei32/aitalked) crate to interface with `aitalked.dll`, allowing speech generation without GUI interaction.

- 🔊 **Simple API Design**  

  Offers a minimal HTTP API for generating speech. *Note: accent position editing is not supported due to API constraints.*

- 📦 **WAV File Output**  

  Returns synthesized speech as a WAV file in the HTTP response.

- 🧠 **Character Info Extraction**  

  Includes a feature to extract character infos from `info.bin`.

## Use Case

Primarily intended for integration with Discord bots or automation tools requiring Japanese TTS capabilities in a headless environment.

## API Details

### `GET /api/voices`

This endpoint returns a list of available voice profiles that can be used with the TTS engine.

#### Response

The response is a JSON array of voice objects. Each object contains:

- `id` *(string)*: A unique identifier for the voice. This is used as the `voice_id` in the `/api/tts` request.

- `name` *(string)*: The display name of the character or voice.

- `icon` *(string)*: A base64-encoded PNG image representing the character's icon.

- `dialect` *(string)*: Describes the regional dialect used by the voice (e.g., "Standard", "Kansai").

- `gender` *(string)*: Indicates the gender of the voice (e.g., "Male", "Female").

- `background_color` *(string)*: A hex color code representing the character's theme or UI background color.

This endpoint is useful for dynamically populating voice selection UIs or validating available options before making synthesis requests.

### `POST /api/tts`

This endpoint generates speech audio from the provided text and voice parameters. It returns a WAV file upon success or a plain-text error message if the request is invalid.

#### Request

The request must be a JSON object with the following fields:

- `voice_id` *(string)*: The identifier of the voice character to use. This should match one of the IDs returned by the `/api/voices` endpoint.

- `text` *(string)*: The input text to be synthesized into speech.

- `is_kansai` *(boolean)* *(optional)*: If set to `true`, the generated speech will use Kansai dialect.

- `volume` *(number)* *(optional)*: Controls the loudness of the voice. Typically ranges from 0 to 1 (maximum 5).

- `speed` *(number)* *(optional)*: Adjusts the speaking rate. Lower values slow down the speech, higher values speed it up.

- `pitch` *(number)* *(optional)*: Modifies the pitch of the voice. Useful for making the voice sound higher or deeper.

- `range` *(number)* *(optional)*: Controls the pitch variation. A higher range adds more expressiveness.

- `pause_middle` *(number)* *(optional)*: Sets the pause duration after commas or mid-sentence breaks.

- `pause_long` *(number)* *(optional)*: Sets the pause duration after long breaks, such as semicolons.

- `pause_sentence` *(number)* *(optional)*: Sets the pause duration at the end of sentences.

#### Response

- `200 OK`: Returns a WAV file containing the synthesized speech.

- `400 BAD_REQUEST`: Returns a plain-text error message describing the issue (e.g., missing fields, invalid values).