ToolNet-SRT

Audio & Video Subtitle Transcription Toolkit

ToolNet-SRT is an open-source toolkit for automatically transcribing audio and video into subtitle files, powered by JianYing (CapCut) speech recognition.

Features

• Automatic audio and video transcription
• Chinese speech recognition
• SRT, ASS, JSON and TXT output
• Word-level timestamps in JSON
• TypeScript SDK and command-line interface (CLI)
• FFmpeg-based audio extraction
• No API key required for the JianYing ASR engine

Requirements

• Node.js 18+
• FFmpeg and FFprobe
• Internet connectivity

Installation from npm

Install the CLI globally:

npm install -g toolnet-srt
toolnet-srt --help

Or install the package in a project to use the SDK:

npm install toolnet-srt

You can also run the CLI without a global installation:

npx toolnet-srt --help

Installation from source

git clone https://github.com/ToolNetWorks/Toolnet-SRT.git
cd Toolnet-SRT
npm ci
npm run build

CLI usage

After installing globally, use the CLI to transcribe audio or video:

toolnet-srt input.mp4 -o output.srt
toolnet-srt input.mp3 -o output.json
toolnet-srt input.wav -o output.ass

When running from a source checkout, replace toolnet-srt with node bin/jianying-subtitle.mjs after building.

TypeScript SDK

After installing the package:

import { transcribe, toSrt } from "toolnet-srt";

const segments = await transcribe({ input: "audio.mp3" });
const srt = toSrt(segments);
console.log(srt);

Render HTTP API

The hosted API is available at https://toolnet-srt.onrender.com. Check its health with:

curl -fsS https://toolnet-srt.onrender.com/api/health

Set your API token locally, then send the audio as the raw request body:

read -rsp "API token: " TOOLNET_SRT_API_KEY
echo

curl --fail-with-body --show-error --max-time 420 \
  -H "X-API-Key: $TOOLNET_SRT_API_KEY" \
  -H "Content-Type: audio/mpeg" \
  --data-binary @audio.mp3 \
  -o subtitle.srt \
  https://toolnet-srt.onrender.com/v1/transcribe

unset TOOLNET_SRT_API_KEY

Use the matching content type for the input: video/mp4, audio/mpeg, audio/wav, audio/x-wav, audio/mp4, or audio/flac. Uploads are limited to 40 MiB. The service accepts one transcription at a time and may return HTTP 429 while busy. The token is required for transcription, but not for the health endpoint. Keep it private.

When hosting the service yourself, set the TOOLNET_SRT_API_KEY environment variable on the server to the token clients should send in X-API-Key.

Development

npm ci
npm run build
npm test

Roadmap

• Long-video chunk processing
• Subtitle merging and timestamp correction
• Optional subtitle translation
• Render deployment
• External application integration

These roadmap features are not yet guaranteed to be available in the current release.

Privacy

Audio is uploaded to ByteDance/JianYing servers for speech recognition. Do not submit confidential audio or video.

License

GNU General Public License v3.0 only. See LICENSE for the complete license text and NOTICE.md for attribution.

Copyright (C) 2026 xifan2333. ToolNetWorks retains copyright in its own original contributions.