Izwi is a voice AI runtime designed for organizations that require private, local processing of audio data. It addresses the needs of teams who cannot send audio to the cloud due to privacy, latency, cost, or policy constraints. The platform enables speech recognition, text-to-speech conversion, speaker diarization, and the orchestration of voice workflows entirely within the user's chosen environment, ensuring that audio, models, and requests remain local.
The tool supports automatic speech recognition (ASR), text to speech (TTS), and speaker diarization (DIAR), allowing users to perform transcription, generate voice output, and separate speakers in meetings or interviews without moving data outside their controlled environment. Izwi also includes a workflow engine (FLOW) that lets teams combine speech services, select models, and integrate application logic behind a single runtime. Integration is streamlined through OpenAI-compatible /v1 API endpoints, making it familiar for teams already using OpenAI-style APIs. This local-first approach allows products and internal platforms to embed private speech features without relying on external cloud voice APIs.
Izwi is suitable for software vendors embedding speech features into regulated products, platform teams seeking unified and governed speech workflows, and edge or offline teams operating in environments with unreliable or restricted connectivity. The runtime can be evaluated locally on a laptop, with a desktop app available for testing and prototyping. Audio can be sourced from a microphone, file, or application, and all processing is handled within the local environment using bundled models and runtime controls.
The platform is open source for evaluation purposes, allowing teams to test and prove workflows before moving to production. For production deployments, Izwi offers enterprise packaging with supported builds, version pinning, observability, and rollout guidance. This commercial path enables organizations to start free and transition to a supported deployment as their needs evolve. Izwi positions itself as voice infrastructure for teams that require predictable control over audio handling, rather than as a generic hosted voice service.
Izwi Ai is a Voice, TTS & speech product. It focuses on enabling private, local deployment of voice AI features like speech recognition and TTS without sending audio to the cloud. Izwi Ai is a B2B product aimed at enterprises and privacy-focused teams. The product follows a commercial open-source model under the MIT license. The product ships for the web, macOS, Windows, Linux, and API, and it can be self-hosted.
Izwi Ai first shipped in 2026. The project is developed in the open on GitHub with 343 stars and 688 commits in the last 90 days. Among its 7 catalogued features are speech recognition, text to speech, and speaker diarization. It exposes integrations via a public API.
Latest indexed changes and source events
Other apps tracked under the same category.
