NarrateAI

Stop recording your voice.

CommunityOnlineAPI key requiredGlobalFreeRead-only

What it can do

  • Get Job Result: Check job status and result. Poll every 60 seconds — do NOT poll more frequently. Video processing typically takes 3-5 minutes. Progress may stay at 20% during frame analysis for 1-3 m
  • Get Upload Url: GET A SIGNED UPLOAD URL for uploading a local video to NarrateAI cloud storage. Use this ONLY when running in HTTP/remote mode and the user has a local video file. After getting the UR
  • Generate Narration Script: NARRATION SCRIPT – generates an AI-written timed script for a SILENT video. No audio output. Use when the user wants a timed narration script, text-only narration, or sync d

What data it sees

Do you need an account

An API key from the service settings is required

Stop recording your voice. NarrateAI adds professional voiceover to silent screen recordings — automatically. Just drop a video URL and get a narrated demo, dubbed video, or polished document back. Works from Cursor, Claude, ChatGPT, or any MCP client.

What it does:

  • Narrates silent videos with AI voiceover — no mic, no script writing, no editing
  • 7 AI voices + voice cloning from a 15-second audio sample
  • Transcribes speech-to-text from any video (meetings, podcasts, tutorials)
  • Translates and dubs videos into any language with cloned speaker voice
  • Converts screen recordings into professional docs (5 templates: onboarding guides, tutorials, feature showcases, product docs, business overviews)
  • Text-to-speech for any text

18 MCP tools — single videos, batch processing (up to 5), transcript editing, voice cloning, TTS, and more.

Free tier: 5 minutes of video processing. Upgrade anytime at narrateai.app.

How to connect

How to connect NarrateAI to Claude

The server address to paste into Claude settings:

https://video--narrateai.run.tools

  1. Open Claude (claude.ai or the desktop app).
  2. Go to Settings → Connectors.
  3. Click Add custom connector.
  4. Paste the server address copied below into Remote MCP server URL and click Add.
  5. In a chat, click + → Connectors and switch the new connector on.

Custom connectors are available on Free, Pro, Max, Team and Enterprise plans (Free is limited to one). On Team and Enterprise an organization Owner adds the connector first under Organization settings → Connectors.

Authorization

After you click Add, a sign-in window for the service opens. Sign in with your own account and approve access. Claude never sees your password.

How to connect NarrateAI to ChatGPT

The server address to paste into ChatGPT settings:

https://video--narrateai.run.tools

  1. Open ChatGPT in a browser (chatgpt.com). A Plus, Pro, Business, Enterprise or Edu plan is required.
  2. Turn on developer mode once: Settings → Apps → Advanced settings → Developer mode.
  3. Open Settings → Connectors and click Create.
  4. Fill in the form: Name (anything), Description (one line about what the service does), MCP server URL (copy it below).
  5. Under Authentication choose OAuth if the service requires sign-in, otherwise None. Click Create.
  6. In a new chat open + → Apps/Connectors and enable the connector.

OpenAI has renamed this section before (Connectors → Apps/Plugins). If the label differs, search settings for "developer mode". On Business/Enterprise workspaces an admin must allow custom connectors first.

Authorization

On first use ChatGPT opens the service's sign-in window. Sign in and approve access.

How to connect NarrateAI to Cursor

The server address to paste into Cursor settings:

https://video--narrateai.run.tools

Fastest: click Open in Cursor below and confirm the prompt.

Manually:

  1. In Cursor open Settings → Cursor Settings → MCP and click Add new global MCP server.
  2. Paste the JSON copied below into ~/.cursor/mcp.json (per project: .cursor/mcp.json in the repo root).
  3. Save the file. The server appears in the MCP list; authorize it there if asked.

Authorization

If the service needs sign-in, an authorize button appears next to the server in the MCP list.

Server tool list (18)

Raw names from tools/list. Only developers need these.

get_job_resultCheck job status and result. Poll every 60 seconds — do NOT poll more frequently. Video processing typically takes 3-5 minutes. Progress may stay at 20% during frame analysis for 1-3 minutes — this is completely normal. Do NOT interpret slow progress as failure. Only report failure when status is "failed" with an error message. Returns: status (processing | transcript_ready | completed | failed | poll_error). poll_error means a temporary connection issue — the job is still running, just retry. transcript_ready includes transcript. completed includes video_url.
get_upload_urlGET A SIGNED UPLOAD URL for uploading a local video to NarrateAI cloud storage. Use this ONLY when running in HTTP/remote mode and the user has a local video file. After getting the URL, upload the file with curl, then pass the returned temp_file_path to any processing tool as the video_source. For stdio/local mode this is NOT needed — tools can read local files directly. Returns: JSON with upload_url, temp_file_path, and a ready-to-use curl command.
generate_narration_scriptNARRATION SCRIPT – generates an AI-written timed script for a SILENT video. No audio output. Use when the user wants a timed narration script, text-only narration, or sync data for a silent video. This does NOT extract existing speech (use transcribe_video for that). This does NOT produce a video file (use narrate_video_full for that). Runs as a background task with progress reporting. Processing takes 1-5 minutes. Returns: JSON with transcript, job_id, db_job_id.
narrate_video_fullFULL NARRATED VIDEO – produces a downloadable video with AI voiceover. Use when the user wants: "narrate this video", "add voiceover", "make a narrated video". VOICE OPTIONS — ask the user which they prefer: 1. AI voice: male1 (default, fastest), female1 (default, fastest), female2, female3, female4, male2, male3 2. Voice cloning: user provides an audio file (voice_sample) and their voice is cloned for the narration If voice_sample is provided, it takes priority over voice_type. Runs as a background task with progress reporting. Processing takes 2-5 minutes. Returns: JSON with video_url, transcript, job_id when done.
abandon_jobAbandon/cancel a processing job. Call this when the user cancels on the agent side. Stops the backend from continuing audio generation and video assembly. Use after narrate_video_transcript or when continue_to_full_video was started but user cancelled. Returns: JSON with success or error.
transcribe_videoTRANSCRIPTION ONLY – video with existing voice -> speech-to-text -> timed transcript. No translation, no narrated video. Returns original speech as-is. Use when the user wants to transcribe a video that already has spoken audio (podcast, interview, meeting recording, etc.). CRITICAL: source_language is REQUIRED. If the user does not specify the language of the video, you MUST ask them concisely before calling. Supported: en, zh, yue, fr, de, it, ja, ko, pt, ru, es (Qwen) + others via Whisper. Runs as a background task with progress reporting. Processing takes 1-5 minutes. Returns: JSON with transcript, job_id, db_job_id when done.
transcribe_and_translateTRANSCRIBE & TRANSLATE (new upload) – video with voice -> speech-to-text -> translate -> translated transcript. No TTS, no video output. Returns translated timed transcript only. Use when the user uploads a new video and wants a translated transcript (e.g. Spanish podcast -> English transcript). CRITICAL: source_language and target_language are REQUIRED. Ask user if not specified. Runs as a background task with progress reporting. Processing takes 1-5 minutes.
translate_existing_videoTRANSLATION (existing video) – Translate transcript of a video already in the user's library. Loads transcript from cloud, translates, returns. No upload. Sync – returns immediately. Use when the user wants to translate a video they already narrated/dubbed with NarrateAI (e.g. "Translate my video X to French"). job_id is the completed video's job ID. CRITICAL: source_language and target_language are REQUIRED. For narrated videos, source is typically the narration language (e.g. en). For dubbed videos, source is the dubbed language.
dub_video_fullFULL AUTO-DUBBING – transcribe -> translate -> extract speaker voice -> TTS with cloned voice -> dubbed video. No refinement screen. Uses the video's own speaker voice for the dubbed audio. Use when the user wants a complete dubbed video (e.g. Spanish video -> English dubbed). CRITICAL: source_language, target_language, and preserve_background_music are REQUIRED. Agent MUST ask user for all three if not specified. For preserve_background_music: ask if the video has background music they want to keep (true) or replace with silence (false). Runs as a background task with progress reporting. Processing takes 2-5 minutes.
generate_documentDOCUMENT GENERATION – produces a structured markdown document from a silent video. Use when the user wants: a document, article, guide, tutorial, or written content based on a video. NOT for narrated video or voiceover. The agent MUST ask which document type the user wants before calling: - user_onboarding: Step-by-step onboarding guide - tutorial_guide: Tutorial/how-to guide - feature_showcase: Feature showcase document - business_overview: Business overview document - product_documentation: Product documentation Also returns a synced transcript as a bonus – offer it to the user after the document is delivered ("I also have a synced transcript for this video, would you like it?"). Runs as a background task with progress reporting. Processing takes 1-5 minutes. Returns: JSON with document_markdown, document_data, transcript, job_id, db_job_id.
generate_ttsTEXT-TO-SPEECH – generate audio from text. Returns a downloadable audio URL. Use when the user wants: "read this aloud", "generate speech", "text to speech", "convert text to audio", "make an audio file from this text". VOICE OPTIONS — ask the user which they prefer: 1. AI voice: male1 (default, fastest), female1 (default, fastest), female2, female3, female4, male2, male3 2. Voice cloning: user provides an audio file (voice_sample) and their voice is cloned If voice_sample is provided, it takes priority over voice_type. Returns: JSON with audio_url, text, voice, language.
narrate_batchBATCH NARRATION – narrate multiple videos in parallel. Each gets a full narrated video with voiceover. Use when the user has multiple videos to narrate (e.g. "narrate these 3 videos"). Maximum 5 videos per batch. Each video is processed independently – one failure does not affect others. If the user does not specify a voice, ask them ONCE (applies to all videos). Voice options: male1 (default, fastest), female1 (default, fastest), female2, female3, female4, male2, male3. CRITICAL – Context handling: Before calling, you MUST ask the user about context: 1. "Would you like to provide the same context for all videos, different context per video, or no context?" 2. If same for all: use manual_context. 3. If different: use contexts_json. 4. If none: leave both empty. Runs as a background task. Processing takes 2-10 minutes depending on video count and length. Returns: JSON array of results with video_url per video.
batch_generate_scriptsBATCH SCRIPT GENERATION – generate AI narration scripts for multiple silent videos in parallel. Each video gets a timed narration script (text only, no audio). Maximum 5 videos per batch. One failure does not affect others. CRITICAL – Context handling: Before calling, ask the user about context: 1. Same for all -> manual_context. 2. Different per video -> contexts_json. 3. No context -> leave both empty. Runs as a background task. Processing takes 2-10 minutes.
batch_transcribeBATCH TRANSCRIPTION – transcribe speech from multiple videos in parallel. Each video must have existing spoken audio. Returns timed transcript per video. CRITICAL: source_language is REQUIRED – ask user if not specified. Applies to all videos. Maximum 5 videos per batch. One failure does not affect others. Runs as a background task. Processing takes 2-10 minutes.
batch_dubBATCH DUBBING – dub multiple videos into another language in parallel. Each video gets full auto-dubbing (transcribe -> translate -> voice clone -> dubbed video). CRITICAL: source_language, target_language, preserve_background_music are REQUIRED – ask user. All videos share the same languages and music setting. Maximum 5 videos per batch. One failure does not affect others. Runs as a background task. Processing takes 2-10 minutes.
update_transcriptUPDATE TRANSCRIPT – edit the narration script before continuing to full video. Use after generate_narration_script returns a transcript and the user wants to change wording, timing, or content of specific segments. The user describes changes naturally; you apply them and call this tool with the updated segments. Also used in the translate-then-re-narrate flow: after translate_existing_video returns a translated transcript, call this with the translated segments, reset_for_reprocessing=True, and target_language set to the translation language (e.g. "hi", "fr") to prepare the completed job for re-narration via continue_to_full_video. The transcript_json must include ALL segments (not just changed ones) — it replaces the full transcript. Each segment needs: start_time, end_time, text. Optionally: pause_duration, chunk_type. After updating, the user can call continue_to_full_video with the same job_id. Returns: JSON with success status or error.
list_videosLIST VIDEOS – get the user's video library (previously processed videos). Use when the user wants to see their existing videos, re-translate a previously narrated video, or work with videos they already processed. Returns paginated list with job IDs, filenames, status, and timestamps. The returned job IDs can be used with translate_existing_video to translate a completed video's transcript, or with get_job_result to check status. Returns: JSON with jobs array (id, filename, status, language, created_at, updated_at), total count, page, and per_page.
continue_to_full_videoContinue from transcript to full narrated video. Use after generate_narration_script returns a transcript and the user is satisfied with it. VOICE OPTIONS — ask the user which they prefer: 1. AI voice: male1 (default, fastest), female1 (default, fastest), female2, female3, female4, male2, male3 2. Voice cloning: user provides an audio file (voice_sample) and their voice is cloned for the narration If voice_sample is provided, it takes priority over voice_type. Runs as a background task with progress reporting. Processing takes 1-3 minutes. Returns: JSON with video_url, job_id when done.