tiktok-transcript-mcp

# TikTok Transcript Scraper — MCP server

Remote MCP server that transcribes public TikTok videos with an AI speech model.
It listens to the audio rather than copying TikTok's auto-captions, so it works on
videos that have no captions at all and across the 90+ languages TikTok creators
actually post in.

**Endpoint:** `https://mcp.apify.com?tools=deapi/tiktok-transcript-scraper`
**Transport:** `streamable-http`
**Actor page:** https://apify.com/deapi/tiktok-transcript-scraper

## What it returns

Per video: clean punctuated text, sentence- or word-level timings, speaker labels
for duets and interviews, ready-to-use SRT and VTT subtitle files, per-segment
confidence scores, and optional TikTok metadata (author, hashtags, views, likes).
Private, deleted and region-locked videos come back as failed rows and are not
charged.

## Authentication

The endpoint **requires authentication** — an unauthenticated request is rejected
with `401 Unauthorized` before the MCP handshake, so there is no anonymous mode.
Authenticate with either:

- **OAuth** — your MCP client opens a browser to sign in to Apify, or
- **Bearer token** — send `Authorization: Bearer <APIFY_TOKEN>`, using a token from
  Apify Console → Settings → API & Integrations.

## Client configuration

```json
{
  "mcpServers": {
    "tiktok-transcript": {
      "url": "https://mcp.apify.com?tools=deapi/tiktok-transcript-scraper",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}
```

Replace `YOUR_APIFY_TOKEN` with your own token. Never commit it.

## Registry

This repository exists to back the server's entry in the
[Official MCP Registry](https://github.com/modelcontextprotocol/registry) under the
namespace `io.github.deapi-ai`. The manifest is [`server.json`](./server.json).