Which integrations are available for this server?

Provides text-to-speech synthesis via Amazon Polly with 100+ voices, 41 languages, and automatic engine selection (neural, standard, generative, long-form). Provides high-quality text-to-speech synthesis with 70+ languages and over 5,000 voices, including voice settings and custom voice IDs. Provides text-to-speech synthesis using OpenAI TTS models with 9 voices and 57 languages, supporting models tts-1 and tts-1-hd.

How do I use langlearn-tts?

1. Click on "Deploy Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@langlearn-tts Pronounce 'bonjour' in French" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

langlearn-tts

by punt-labs

Overview Schema Source Related Servers Discussions Admin

Python

Local

langlearn-tts

A Claude Desktop extension that gives Claude the ability to speak. Ask Claude to pronounce words, generate audio flashcards, or run a full language lesson with audio — in 70+ languages.

Status (2026-02-21)

Core CLI and MCP surfaces are working with ElevenLabs, OpenAI TTS, and AWS Polly.
Provider defaults and audio variants (word + example) are still being standardized.
Claude Desktop directory submission is pending external review.

Related MCP server: elevenlabs-voice-mcp

Roadmap

See ROADMAP.md.

Quick Start

1. Get a TTS API key

You need an account with at least one text-to-speech provider:

ElevenLabs — best quality, 70+ languages, 5,000+ voices. Free tier: 10K chars/month. (Recommended)
OpenAI TTS — good quality, easiest setup, 57 languages, 9 voices.
AWS Polly — better quality, 41 languages, 100+ voices, difficult setup (setup guide).

2. Install in Claude Desktop

Download punt-langlearn-tts.mcpb and double-click to install. Claude Desktop will prompt you for your API key and an output directory.

3. Set up a tutor project (optional)

langlearn-tts ships with 28 tutor prompts — one for each combination of 7 languages and 4 levels. Setting up a project gives Claude a tutor persona that generates audio during lessons.

In Claude Desktop, click Projects in the sidebar
Click Create Project and name it (e.g., "German with Herr Schmidt")
Open the project, click Set custom instructions
Copy a prompt from the prompts directory and paste it into the Instructions field
Start a new conversation within that project

Language	High School	1st Year	2nd Year	Advanced
German	Herr Schmidt	Professorin Weber	Professor Hartmann	Professor Becker
Spanish	Profesora Elena	Profesor Garcia	Profesora Carmen	Profesora Reyes
French	Madame Moreau	Professeur Laurent	Professeur Dubois	Professeur Beaumont
Russian	Irina Petrovna	Professor Dmitri	Professor Natasha	Professor Mikhail
Korean	Kim-seonsaengnim	Professor Park	Professor Kim	Professor Yoon
Japanese	Tanaka-sensei	Yamamoto-sensei	Suzuki-sensei	Mori-sensei
Chinese	Laoshi Wang	Professor Chen	Professor Zhang	Professor Wei

Each prompt is calibrated to the student's level, based on Mollick & Mollick's "Assigning AI" framework.

4. Try it out

In any Claude Desktop conversation, try:

"Say 'Guten Morgen' in German"
"Create an audio flashcard: 'good morning' in English, then 'Guten Morgen' in German"
"Synthesize these Spanish words as a merged audio file: hola, gracias, por favor, de nada"
"Generate pair flashcards for these German vocabulary words: strong/stark, house/Haus, book/Buch"

Audio plays automatically after each request. Files are saved to your output directory (~/langlearn-audio by default).

Features

Pronounce anything — ask Claude to say a word or phrase and hear it spoken aloud
Audio flashcards — Claude creates an MP3 with English first, then the target language, with a pause between them
Vocabulary lists — give Claude a list of words and get back individual or merged audio files
70+ languages — German, Spanish, French, Russian, Korean, Japanese, Chinese, and many more
Tutor mode — 28 built-in tutor personas that teach with audio throughout the lesson
Multiple voices — each provider offers a range of voices; ask Claude to use a specific one by name
Adjustable speed — audio defaults to 90% speed so learners can hear pronunciation clearly

Troubleshooting

If something isn't working, ask Claude to run a health check:

"Run the doctor command to check if everything is set up correctly"

Logs are written to ~/.langlearn-tts/logs/langlearn-tts.log (never contains the text you synthesize). See PRIVACY.md for details.

Developer Reference

Everything below is for developers using the CLI, integrating with other MCP clients, or contributing to the project.

Claude Code / CLI

curl -fsSL https://raw.githubusercontent.com/punt-labs/langlearn-tts/14f4194/install.sh | sh

The default provider is AWS Polly. To use a different provider:

LANGLEARN_TTS_PROVIDER=elevenlabs curl -fsSL https://raw.githubusercontent.com/punt-labs/langlearn-tts/14f4194/install.sh | sh

uv tool install punt-langlearn-tts
langlearn-tts install --provider polly
langlearn-tts doctor

curl -fsSL https://raw.githubusercontent.com/punt-labs/langlearn-tts/14f4194/install.sh -o install.sh
shasum -a 256 install.sh
cat install.sh
sh install.sh

Install ffmpeg for audio stitching (pairs, merged batches):

# macOS (requires Homebrew — install from https://brew.sh if needed)
brew install ffmpeg

# Linux — see https://ffmpeg.org/download.html for your distro

# Windows
winget install --id Gyan.FFmpeg

Claude Desktop setup via CLI

langlearn-tts install

Writes to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS). Options: --provider NAME, --output-dir PATH, --uvx-path PATH. Restart Claude Desktop after running.

Or add manually:

{
  "mcpServers": {
    "langlearn-tts": {
      "command": "/absolute/path/to/uvx",
      "args": ["--from", "punt-langlearn-tts", "langlearn-tts-server"],
      "env": {
        "LANGLEARN_TTS_OUTPUT_DIR": "/absolute/path/to/output/directory"
      }
    }
  }
}

Claude Desktop does not inherit your shell environment. API keys must be literal values (env var references are not supported). Restart after editing.

Environment variables

Env var	Required	Description
`LANGLEARN_TTS_PROVIDER`	No	`elevenlabs`, `polly` (default when no API key), or `openai`
`ELEVENLABS_API_KEY`	For ElevenLabs	Your API key
`OPENAI_API_KEY`	For OpenAI	Your API key
`LANGLEARN_TTS_OUTPUT_DIR`	No	Output directory (default: `~/langlearn-audio`)
`LANGLEARN_TTS_MODEL`	No	Model name. ElevenLabs: `eleven_v3` (default). OpenAI: `tts-1`, `tts-1-hd`

For Polly, AWS credentials are read from ~/.aws/credentials.

CLI Usage

# Single synthesis
langlearn-tts synthesize "Guten Morgen" --voice daniel -o morning.mp3

# Custom speech rate (percentage, default 90)
langlearn-tts synthesize "Привет" --voice tatyana --rate 70 -o privet.mp3

# ElevenLabs with voice settings
langlearn-tts synthesize "Guten Morgen" --voice Rachel \
  --stability 0.5 --similarity 0.7 --style 0.3 --speaker-boost

# Pair: English + German stitched with a pause
langlearn-tts synthesize-pair "good morning" "Guten Morgen" \
  --voice1 joanna --voice2 daniel -o pair.mp3

# Batch from JSON file (["hello", "world", "good morning"])
langlearn-tts synthesize-batch words.json -d output/

# Batch merged into single file
langlearn-tts synthesize-batch words.json -d output/ --merge --pause 800

# Pair batch from JSON file ([["strong", "stark"], ["house", "Haus"]])
langlearn-tts synthesize-pair-batch pairs.json -d output/

# Browse AI tutor prompts
langlearn-tts prompt list
langlearn-tts prompt show german-high-school | pbcopy

Voices

ElevenLabs — 5,000+ voices. Any voice works with any language. You can also pass a voice ID directly (the 20-character string from the ElevenLabs dashboard). Voice settings: --stability, --similarity, --style (0.0–1.0), --speaker-boost (flag).

AWS Polly — 93 voices from the AWS Polly voice list. Each voice is trained for a specific language. Engine (neural, standard, generative, long-form) is selected automatically.

OpenAI TTS — 9 voices: alloy, ash, coral, echo, fable, onyx, nova, sage, shimmer. Default model: tts-1. Use --model tts-1-hd for higher quality.

All voice names are case-insensitive.

MCP Tools

Tool	Description
`synthesize`	Single text to MP3
`synthesize_batch`	Multiple texts, optionally merged
`synthesize_pair`	Two texts stitched with a pause
`synthesize_pair_batch`	Multiple pairs, optionally merged

Each tool accepts auto_play (default: true) to play audio immediately after synthesis.

Other MCP clients

langlearn-tts works with any MCP client that supports stdio transport. Use the server command uvx --from punt-langlearn-tts langlearn-tts-server with the environment variables above. Find your uvx path with which uvx — all paths must be absolute.

Development

git clone https://github.com/punt-labs/langlearn-tts.git
cd langlearn-tts
uv sync --all-extras

uv run pytest tests/ -v
uv run ruff check src/ tests/
uv run ruff format src/ tests/
uv run mypy src/ tests/
uv run pyright src/ tests/

License

MIT

Available Tools

4 tools

synthesizeA

Synthesize text to an MP3 audio file.

Args: text: The text to convert to speech. With ElevenLabs eleven_v3, you can embed audio tags in square brackets anywhere in the text to control delivery — e.g. [tired], [excited], [whisper], [sad], [sigh], [laughs], [dramatic tone]. Tags are free-form; the model interprets them as performance cues. Combine with punctuation (ellipsis for pauses, ! for emphasis) for best results. Tags only work with ElevenLabs eleven_v3 model. voice: Voice name. Default: provider's default voice (currently matilda for ElevenLabs, joanna for Polly, nova for OpenAI). If language is provided without voice, a suitable default voice for that language is selected automatically. language: ISO 639-1 language code (e.g. 'de', 'ko', 'fr'). Enables language-aware voice selection and validation. With Polly, validates voice-language compatibility. With ElevenLabs/OpenAI, passed through (voices are multilingual). rate: Speech rate as percentage (90 = 90% speed, good for language learners). Defaults to 90. ElevenLabs ignores rate; use audio tags like [rushed] or [drawn out] instead. auto_play: Open the file in the default audio player after synthesis. Defaults to true. output_path: Full path for the output file. If not provided, a file is auto-generated in output_dir. output_dir: Directory for output. Defaults to TTS_OUTPUT_DIR env var or ~/langlearn-audio/. stability: ElevenLabs voice stability (0.0-1.0). Ignored by other providers. Defaults to provider default. similarity: ElevenLabs voice similarity boost (0.0-1.0). Ignored by other providers. Defaults to provider default. style: ElevenLabs voice style/expressiveness (0.0-1.0). Ignored by other providers. Defaults to provider default. speaker_boost: ElevenLabs speaker boost toggle. Ignored by other providers. Defaults to provider default.

Returns: JSON string with path, text, voice, and language fields.

Inputs: text (required), rate, style, voice, +7 more. Returns: result

synthesize_batchA

Synthesize multiple texts to MP3 files.

Args: texts: List of texts to synthesize. With ElevenLabs eleven_v3, embed audio tags like [tired], [excited], [whisper] in text. voice: Voice name for all texts. Default: provider's default voice (currently matilda for ElevenLabs, joanna for Polly, nova for OpenAI). If language is provided without voice, auto-selects. language: ISO 639-1 language code (e.g. 'de', 'ko'). rate: Speech rate as percentage. Defaults to 90. merge: If true, produce one merged file instead of separate files per text. Defaults to false. pause_ms: Pause between segments in milliseconds when merging. Defaults to 500. auto_play: Open the file(s) in the default audio player after synthesis. Defaults to true. output_dir: Directory for output files. Defaults to TTS_OUTPUT_DIR env var or ~/langlearn-audio/. stability: ElevenLabs voice stability (0.0-1.0). similarity: ElevenLabs voice similarity boost (0.0-1.0). style: ElevenLabs voice style/expressiveness (0.0-1.0). speaker_boost: ElevenLabs speaker boost toggle.

Returns: JSON string with list of results, each containing path, text, voice, and language fields.

Inputs: texts (required), rate, merge, style, +8 more. Returns: result

synthesize_pairA

Synthesize a pair of texts and stitch them into one MP3.

Creates [text1 audio] [pause] [text2 audio]. Use for language learning pairs like "strong" (English) + "stark" (German).

Args: text1: First text (typically English). With ElevenLabs eleven_v3, embed audio tags like [tired], [excited], [whisper] in text. text2: Second text (typically target language). Same audio tag support as text1. voice1: Voice for text1. Defaults to provider's default voice. If lang1 is provided without voice1, auto-selects. voice2: Voice for text2. Defaults to provider's default voice. If lang2 is provided without voice2, auto-selects. lang1: ISO 639-1 language code for text1 (e.g. 'en'). lang2: ISO 639-1 language code for text2 (e.g. 'de'). rate: Speech rate as percentage. Defaults to 90. pause_ms: Pause between the two texts in milliseconds. Defaults to 500. auto_play: Play the audio after synthesis. Defaults to true. output_path: Full path for the output file. output_dir: Directory for output. Defaults to TTS_OUTPUT_DIR env var or ~/langlearn-audio/. stability: ElevenLabs voice stability (0.0-1.0). similarity: ElevenLabs voice similarity boost (0.0-1.0). style: ElevenLabs voice style/expressiveness (0.0-1.0). speaker_boost: ElevenLabs speaker boost toggle.

Returns: JSON string with path, text, voice, and language fields.

Inputs: text1 (required), text2 (required), rate, lang1, +11 more. Returns: result

synthesize_pair_batchA

Synthesize multiple text pairs and stitch each into MP3 files.

Each pair becomes [text1 audio] [pause] [text2 audio]. Use for vocabulary lists like [["strong","stark"], ["house","Haus"]].

Args: pairs: List of [text1, text2] pairs. With ElevenLabs eleven_v3, texts can include audio tags like [tired], [excited]. voice1: Voice for all first texts. Defaults to provider's default. If lang1 is provided without voice1, auto-selects. voice2: Voice for all second texts. Defaults to provider's default. If lang2 is provided without voice2, auto-selects. lang1: ISO 639-1 language code for first texts (e.g. 'en'). lang2: ISO 639-1 language code for second texts (e.g. 'de'). rate: Speech rate as percentage. Defaults to 90. pause_ms: Pause between pair segments in milliseconds. Defaults to 500. merge: If true, produce one merged file instead of separate files per pair. Defaults to false. auto_play: Play the audio after synthesis. Defaults to true. output_dir: Directory for output files. Defaults to TTS_OUTPUT_DIR env var or ~/langlearn-audio/. stability: ElevenLabs voice stability (0.0-1.0). similarity: ElevenLabs voice similarity boost (0.0-1.0). style: ElevenLabs voice style/expressiveness (0.0-1.0). speaker_boost: ElevenLabs speaker boost toggle.

Returns: JSON string with list of results.

Inputs: pairs (required), rate, lang1, lang2, +10 more. Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

Deploy Server

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.3/5.0

Scored 2026-08-30 04:55 across 4 tools

Disambiguation5/5

Each tool has a clearly distinct purpose: single text, batch texts, a single pair, and multiple pairs. The descriptions explicitly differentiate use cases, making confusion unlikely.

Naming Consistency5/5

All tools follow the same verb-based pattern, starting with 'synthesize' and adding descriptive suffixes (batch, pair, pair_batch). The naming is fully consistent and predictable.

Tool Count5/5

Four tools provide a focused yet complete range of synthesis options, from single to batch to pair-based. This is well-scoped for a TTS server without being bloated.

Completeness4/5

The toolset covers single, batch, pair, and pair-batch synthesis comprehensively. The only notable omission is a way to discover available voices or languages without synthesizing, which would aid agents in choosing parameters.

Maintenance

ActivityNo data

ResponsivenessOutdated data

Resources

Related MCP Connectors

UtteraA
ai.uttera
Audio for your agent: transcribe, speak, translate, summarise, plus sound effects and music.
Audio Processing Speech Processing Language Translation
131
https://mcp.uttera.ai/mcp
Presentr Analyze
io.github.michaelw-pr
Read text aloud in 48 languages, transcribe audio, extract and translate PDF text, dub video.
Speech Processing Multimedia Processing Text Summarization
1
https://analyze.presentr.me/api/mcp
Typeshoot
com.typeshoot
Images, video, voiceovers, music and editing inside Claude and ChatGPT, priced in credits.
1
https://typeshoot.com/mcp
transcriptionA
com.scriptivox.www
Transcribe audio and video and work with your transcript library from any AI app. 119 languages.
Speech Processing Audio Processing AI & Machine Learning
19
https://platform.scriptivox.com/mcp

Related MCP Servers

MCP Audio Server
Text-to-Speech Audio Processing Multimedia Processing
BatchLion
A
license
Not graded
quality
D
maintenance
Enables text-to-speech conversion with smart language detection (using Google TTS for Chinese) and audio file playback with playback controls for Claude Desktop.
Updated a year ago (2025-07-11 11:28 UTC)
2
MIT
elevenlabs-voice-mcp
Text-to-Speech Speech Processing
krystiangw
A
license
A
quality
D
maintenance
Enables Claude to speak responses out loud using ElevenLabs text-to-speech API, with tools to list voices and convert text to speech.
Updated 5 months ago (2026-06-19 09:02 UTC)
2
MIT
muse-tts-embed
Text-to-Speech Speech Processing Audio Processing
falcoschaefer99-eng
F
license
Not graded
quality
D
maintenance
Enables Claude to speak text with an embedded audio player, supporting 54 voices, voice cloning, and playback controls, all running locally.
Updated 6 months ago (2026-05-23 14:43 UTC)
3
-
speak
Text-to-Speech Speech Processing Audio Processing
CareyScott
A
license
Not graded
quality
B
maintenance
Lets Claude convert its replies into natural spoken scripts and play them with pause, resume, and rewind controls, plus voice and engine selection.
Updated 2 months ago (2026-09-14 12:06 UTC)
11 npm
MIT

langlearn-tts

Status (2026-02-21)

Roadmap

Quick Start

1. Get a TTS API key

2. Install in Claude Desktop

3. Set up a tutor project (optional)

4. Try it out

Features

Troubleshooting

Developer Reference

Claude Code / CLI

Claude Desktop setup via CLI

Environment variables

CLI Usage

Voices

MCP Tools

Other MCP clients

Development

License

Available Tools

Tool Schema Changelog

TDQS

Maintenance

Resources

Related MCP Connectors

Related MCP Servers

MCP Audio Server

elevenlabs-voice-mcp

muse-tts-embed

speak