RagLeapRagLeap Core is open-source software (MIT) that you run yourself. The hosted RagLeap service that used to be reachable at this address has been retired. This page separates what voice support exists in the open-source project from what is only research.
What exists in RagLeap Core
The app includes a voice service and a voice channel that connects through Twilio. The channels documentation explains how it works and what you need to set up.
What ragleap-voice is, and is not
ragleap-voice is a separate package, version 0.0.2, pre-alpha. It is the model-management layer only: a licence-checked manifest of candidate speech models, lock files that record the size and SHA-256 of every file at a pinned revision, and downloads from Hugging Face only (HTTPS, a host allowlist, and the access token is sent only to huggingface.co). Every manifest entry is unverified: a research lead, not an endorsement. There are no speech engines in this release, so you cannot transcribe or synthesize audio with it yet. See the library page.
What has been measured
Published benchmarks use small samples of read speech on one GPU. In them an open Indian-language recogniser (IndicConformer) had clearly lower word error than Whisper large-v3 on Hindi, Tamil, Telugu, Malayalam and Kannada. Naturalness, emotion quality and comparisons with commercial services have not been measured, and this project makes no claim about them. The benchmarks are in the repository.
Planned, not available
A speech-to-text and text-to-speech API with engines, streaming, voice profiles with a consent record, emotion and style control, a registry of community-trained models, and more languages.
Limits you should know
- Tested by the maintainer on one Windows laptop and on Linux servers. macOS, Linux desktops and the Ollama option on Windows are not verified yet.
- No named users or roles yet, and no independent security review yet.