Integration

Palatine Speech integration

Palatine Speech is a Russian speech AI cloud. Transcribe audio and video files (uploaded or by HTTPS link) with timestamps, optionally labelling speakers, naming their roles, tagging the emotion of each turn (Russian speech) and masking profanity; label speakers only; score the sentiment of a recording; summarize a recording as meeting minutes or with your own prompt; ask questions about a finished transcript and read its digest; export subtitles and tables; mask profanity in text; read the account balance. Jobs are billed in rubles per audio minute or per token from the account balance, and transcription, diarization, sentiment and summary jobs are left on the server until finished with the matching finish tool. Not exposed: the OpenAI-compatible synchronous transcription and LLM chat routes, the plain-text summary route (its request body is described two different ways), and the routes the vendor reserves for its web cabinet (transcript editing and the request log). Authentication is an API token from the Palatine Speech account, sent as a Bearer token.

Connect Palatine Speech How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in Palatine Speech.Built from the 23 Palatine Speech tools below, on the connection you add once.

What you can do with Palatine Speech

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Transcription, 5 operations

download_transcriptionDownload transcriptionGETDownloads a finished transcription as a file (srt, vtt, txt, csv, xlsx or json), stored for a later node. Answers 409 while the task is still running and 404 for an unknown task.
finish_transcription_taskFinish transcription taskPOSTMarks a transcription task as finished and removes it from the active tasks. Call it once you have the result and no longer need the task.
get_transcription_taskGet transcription taskGETReturns a transcription task: status (success, pending or error) and, once done, data with text, segments and words. response_format srt or vtt returns the subtitles as data.text instead.
transcribe_fileTranscribe filePOSTTranscribes an uploaded audio or video file (text, segments, words); diarization_model adds speakers. Billed per minute. Then call finish_transcription_task. Waits for the job and returns the finished result; gives up after 10 minutes.
transcribe_urlTranscribe urlPOSTTranscribes an audio or video file fetched from an HTTPS link (text, segments, words). Billed per minute. Then call finish_transcription_task. Waits for the job and returns the finished result; gives up after 10 minutes.

Diarization, 4 operations

diarize_fileDiarize filePOSTLabels who speaks when in an uploaded audio or video file (segments, no text). Billed per minute. Then call finish_diarization_task. Waits for the job and returns the finished result; gives up after 10 minutes.
diarize_urlDiarize urlPOSTLabels who speaks when in an audio or video file fetched from an HTTPS link. Billed per minute. Then call finish_diarization_task. Waits for the job and returns the finished result; gives up after 10 minutes.
finish_diarization_taskFinish diarization taskPOSTMarks a diarization task as finished and deletes its cached data. Call it once you have the result.
get_diarization_taskGet diarization taskGETReturns a diarization task: status (success, pending or error) and, once done, the speaker segments.

Sentiment analysis, 3 operations

analyze_sentiment_fileAnalyze sentiment filePOSTScores speech sentiment in an uploaded audio or video file (Very Negative to Very Positive). Billed per minute. Then call finish_sentiment_task. Waits for the job and returns the finished result; gives up after 10 minutes.
finish_sentiment_taskFinish sentiment taskPOSTMarks a sentiment analysis task as finished and removes it from the active tasks. Call it once you have the result.
get_sentiment_taskGet sentiment taskGETReturns a sentiment analysis task: status (success, pending or error) and, once done, the sentiment classes with probabilities.

Summary, 3 operations

finish_summary_taskFinish summary taskPOSTMarks a summary task as finished and removes it from the active tasks. Call it once you have the result.
get_summary_taskGet summary taskGETReturns a summary task: status (success, pending or error) and, once done, the meeting minutes or the answer to the prompt.
summarize_fileSummarize filePOSTSummarizes an uploaded recording as meeting minutes or per your prompt (see task). Billed per token. Then call finish_summary_task. Waits for the job and returns the finished result; gives up after 10 minutes.

Transcript chat, 3 operations

ask_transcript_questionAsk transcript questionPOSTAsks a question about a finished transcript; billed per LLM use. The answer is written in the background: the question comes back pending, then poll get_transcript_chat until its status is success or error.
clear_transcript_chatClear transcript chatDELETEDeletes every question and answer about a transcript and resets the question allowance. An answer still being written is discarded (and not billed unless it was already being charged).
get_transcript_chatGet transcript chatGETReturns the questions and answers about a transcript and the remaining question allowance. Poll it while the last message is pending. May start building the recording digest if there is none.

Transcript digest, 2 operations

get_transcript_digestGet transcript digestGETReturns the digest of a transcript (sections, decisions, action items, figures, open questions) and a meeting summary. Poll while status is building. Starts the build if there is none yet.
rebuild_transcript_digestRebuild transcript digestPOSTRebuilds the digest of a transcript (billed; parts already paid for the same transcript version are reused), for example after an error. Runs in the background: poll get_transcript_digest.

Balance, 1 operation

get_balanceGet balanceGETReturns the balance of the account that owns the token: paid funds (real_money) and bonus funds (virtual_money). Bonus funds are spent first.

Profanity, 1 operation

filter_profanityFilter profanityPOSTMasks obscene and rude words in a text or in up to 100 texts (100,000 characters per call in total), keeping punctuation and length. Free. Pass text or texts, not both.

Upload limit, 1 operation

get_upload_limitsGet upload limitsGETReturns the upload limits the service applies: file size, audio duration and the size limit for files sent by link. null means no limit.

Use Palatine Speech from Claude Code, Cursor or Codex

Connect Palatine Speech in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its Palatine Speech operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect Palatine Speech

Create your workspace
Sign up free. Nothing to install.
Connect Palatine Speech once
Add your Palatine Speech connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.