Eleven Reaper Lab

Eleven Reaper Lab connects REAPER with ElevenLabs, bringing Text to Speech, Voice Conversion, and Speech to Text into your workflow.

Generate audio from text, convert a selected voice, or transcribe audio and edit the text before generating a new take. The ReaImGui interface provides model-specific controls, searchable voices, configurable profiles, and an activity log.

Developed by Armando Liccardo (geckod22).

This is an independent project, not affiliated with or endorsed by Cockos or ElevenLabs.

Requirements

• REAPER with Python enabled for ReaScript.
• Python with the requests package installed in the environment used by REAPER.
• SWS / S&M Extensions for item management and loudness operations.
• ReaImGui, available through ReaPack.
• An ElevenLabs API key with access to the voices and models you want to use.

Installation

1. Install Python, then open Options > Preferences > Plug-ins > ReaScript in REAPER. Enable Python for ReaScript and configure the Python shared library path for your platform.

   To locate the Python executable, use where python3 on Windows or ls -l $(which python3) on macOS.

2. Install the remaining dependencies listed above.

3. Download and extract the project. Place the entire project folder inside your REAPER Scripts directory:

   • Windows: %AppData%\REAPER\Scripts
   • macOS: ~/Library/Application Support/REAPER/Scripts

4. Before the first launch, copy erl_config.example.json to erl_config.json in the same folder as ERL_App.lua. Customize the copy with the ElevenLabs voice IDs you can access, your profiles, and your preferred model defaults. Keep the example file as a reference.

5. In REAPER, open Actions > Show action list, choose New Action > Load ReaScript..., and select ERL_App.lua from the extracted folder.

6. Run Script: ERL_App.lua and enter your API key under Settings.

The application registers its Python actions automatically on first launch. Keep all supplied scripts, JSON files, and the lib folder together.

Updating: back up and preserve your customized erl_config.json, update the application files together, and restart REAPER to unload cached Python modules. Review changes to erl_config.example.json before manually incorporating any new settings into your configuration.

Usage

Choose an operation tab and a model. For speech generation and conversion, select a voice and profile, then adjust the model's settings as needed. Reset preset restores the configured values. Expand Settings to change the API key or audio normalization preference.

Starting an operation displays a highlighted waiting message and changes the disabled action button to Generating audio..., Converting voice..., or Transcribing audio... before processing begins. The message stays static while REAPER waits; the status shows completion or failure when the request returns.

Text to Speech

Enter your text in the editor and generate audio:

• With one audio item selected: the result is aligned with the source on the TTS track, saved beside the source file, and the original item is muted.
• With no items selected: save the REAPER project first. The result is inserted at the edit cursor on the TTS track and saved in ElevenReaperLab_Audio beside the .rpp file.

The destination is shown before generation. Selecting multiple items is not supported.

Text to Speech does not upload the selected item's audio. The item only provides alignment and the output location, so it does not need Glue and its duration does not limit text generation. With no item selected, generation uses the saved project and edit cursor as described above.

After a successful audio import, only the newly generated item remains selected, including after optional normalization. The edit cursor and track selection are preserved. If importing fails, the previous item selection is restored.

Voice Conversion

Select one audio item and use Voice Conversion to convert it to the chosen voice. The result is inserted on the VOICE_REGEN track and the original item is muted. This operation uses the source audio directly; it does not use the text editor.

Speech to Text

Select one audio item and use Speech to Text to transcribe it. After a successful transcription, the application opens Text to Speech with the transcribed text ready to edit. Audio is generated only when you request it.

The Activity log shows operation results and errors. Expand it to filter, copy, or clear entries. You can also assign a keyboard shortcut to the application in REAPER's Action List.

For Text to Speech, the log also shows the credits used when ElevenLabs returns billing metadata. Otherwise it reports that credit usage is unavailable; it does not estimate a charge. This reporting adds no API requests and does not apply to Voice Conversion or Speech to Text.

Configuration and profiles

erl_config.example.json is the distributed template; erl_config.json is your personal configuration. Create the latter by copying the example, then edit it to manage voices, model defaults, usage profiles, and the default normalization preference. The example profiles are Bedtime, Meditation, and Course. Keep your API key in the application's Settings, outside the JSON file.

After editing, click Reload configuration under Settings. A successful reload reapplies presets and clears manual setting changes. Invalid configuration at startup blocks operations until corrected; a failed reload retains the last valid configuration for the current session.

The application requires a valid erl_config.json; it does not create it automatically or load the example as a fallback. To restore the example settings, back up your personal file before replacing it with a copy of erl_config.example.json. See the Configuration Guide for examples, setting precedence, and validation rules.

Limitations

• REAPER remains blocked during service requests. Live progress and responsive cancellation are unavailable while a request is running.
• Voice Conversion and Speech to Text require one valid audio item, with a maximum duration of 120 seconds. Follow the application's Glue requirements before processing. These Glue and duration restrictions do not apply to Text to Speech.
• Available controls come from the bundled model catalog. Opening or reloading the application does not check which models or voices your ElevenLabs account can access. Realtime transcription is not supported.
• Eleven v3 Stability choices are provisional and have not been verified through the API; the interface displays a compatibility warning.
• Requests use a 10-second connection timeout and a 300-second read-inactivity timeout. They are not retried automatically. If the outcome is uncertain, check the service before repeating a generation to avoid duplicate requests and charges.

Documentation and third-party library

• Configuration Guide: voices, profiles, defaults, and configuration errors.
• Model Registry: registered models, parameter capabilities, and references.
• Protocol Guide: request handling, errors, and recovery behavior.

The bundled rxi/json.lua library requires no separate installation. Its MIT license is included in lib/json.LICENSE.