typecast-api-mcp-server-sample

What is typecast-api-mcp-server-sample?

It is an MCP (Model Context Protocol) server for the Typecast API, enabling MCP clients like Claude Desktop or Cursor to interact with Typecast’s text-to-speech and voice management features in a standardized way.

How to use typecast-api-mcp-server-sample?

Configure it in your MCP client’s claude_desktop_config.json or Cursor MCP settings with the command uvx and set environment variables TYPECAST_API_KEY and TYPECAST_OUTPUT_DIR. Alternatively, clone the repository and run locally with Python 3.10+ and uv.

Key features of typecast-api-mcp-server-sample

• Voice management: list, get, and recommend voices.
• Text-to-speech with standard, streaming, and timestamped output.
• Supports ssfm-v30 model with 7 emotion presets and Smart Mode.
• Quick Voice Cloning from local WAV/MP3 files (≤25 MB).
• Audio playback and loudness normalization via target_lufs.
• Tools available: clone_voice, delete_cloned_voice, text_to_speech, text_to_speech_stream, text_to_speech_with_timestamps.

Use cases of typecast-api-mcp-server-sample

• Generate speech from text using different voices and emotional tones.
• Create temporary custom voices by cloning audio samples for a single session.
• Retrieve voice recommendations based on desired style, mood, language, or use case.
• Integrate Typecast TTS into AI assistant workflows via MCP clients.

FAQ from typecast-api-mcp-server-sample

What emotions are supported?

ssfm-v30 offers 7 presets: normal, happy, sad, angry, whisper, toneup, and tonedown, plus a Smart Mode that infers emotion from context using previous_text and next_text. ssfm-v21 supports preset only.

What are the constraints for Quick Voice Cloning?

Voice name must be 1–30 characters, the audio sample must be WAV or MP3, and the file size must be 25 MB or smaller. Use the ssfm-v30 model unless you have a specific compatibility reason.

How do I set up the server?

Set environment variables TYPECAST_API_KEY (required) and TYPECAST_OUTPUT_DIR (defaults to ~/Downloads/typecast_output). Then add the configuration to your MCP client’s settings. For Linux, also add XDG_RUNTIME_DIR.

Does it require an API key?

Yes, a Typecast API key is required and must be provided as the TYPECAST_API_KEY environment variable.

How many languages does the server support?

The ssfm-v30 model supports 37 languages. The ssfm-v21 model’s language support is not specified in the README.