Pocket Voice enables users to clone voices and generate speech directly within their web browser. Designed to operate fully client-side, it leverages ONNX Runtime and WebAssembly to process all data locally, ensuring that no audio or text leaves the user's machine. This approach removes the need for server interaction or API keys, emphasizing privacy and user control over their data.
The tool allows individuals to record their own voice through a microphone or upload an audio clip, from which it extracts voice embeddings using the mimi encoder. Once a voice is cloned, users can type text and have it spoken in the cloned voice, with audio generated and streamed in real time via AudioWorklet. Pocket Voice also supports saving cloned voices for later reuse, providing a personalized voice library within the browser session. Additionally, it offers built-in voices and language options including English, German, Italian, Portuguese, and Spanish, enabling text-to-speech generation in multiple languages.
A feature called Voice Studio lets users select a preset, type text, and generate speech using either a newly cloned or a saved voice. The process requires only about five seconds of recorded speech or a short uploaded audio clip to create a voice clone. All ONNX models are loaded from the HuggingFace CDN, but inference and processing remain entirely within the browser, maintaining strict data privacy.
Pocket Voice is intended for legitimate, ethical use and cautions against using the tool for voice cloning without consent or for any harmful, illegal, or deceptive purposes. The tool is built by Hemanth HM and is powered by Pocket TTS. Its client-side architecture makes it suitable for users seeking privacy-preserving voice cloning and text-to-speech capabilities without relying on external servers or cloud services.
Pocket Voice is a Voice, TTS & speech product. It focuses on cloning and generating speech from any voice without sending data to external servers, ensuring privacy and local processing. It is built as a consumer product for privacy-conscious users and hobbyists interested in voice cloning or TTS. The product is available for free. Pocket Voice is available on the web.
Hemanth HM builds and maintains Pocket Voice, and the product first shipped in 2026. Development happens publicly on GitHub with 7.7k stars and 14 commits in the last 90 days. Key capabilities include voice cloning, text-to-speech, and client-side inference.
Latest indexed changes and source events
Show HN: Pocket Voice: TTS and Voice clone in the browser verified by the PulseGate indexer
Other apps tracked under the same category.
