Skip to content

Releases

Hush Studio 0.1.17

Hush Studio 0.1.17 Latest
Latest

Choose a tag to compare

View all tags
@orailnoor orailnoor released this 24 Oct 14:36
v0.1.17
e3a3eb3

Fixed

  • On Android 15 and newer, the studio and player screens no longer run under the status bar and navigation bar. The title, Export button and playback controls now stay clear of them.
  • Layout also allows for display cutouts and side insets.

Updating

Install the APK over 0.1.16. Your voices, projects and speech model are kept. If you are installing for the first time, see the 0.1.16 release notes for setup and requirements.

Verify the download

  • SHA-256 of the APK: 10b578adafd54c67d9daac333d8c11648eeca32bd24e192e0c1a33a115925d19
  • Signing certificate SHA-256: 2c36404b24fdb95fc7e75f86063e3ee4df5c0ccf67936f576ccafe869562e25a (same as 0.1.16)
Assets 4
Loading

Hush Studio 0.1.16

Hush Studio 0.1.16

Choose a tag to compare

View all tags
@orailnoor orailnoor released this 24 Oct 13:39
v0.1.16
48e0d02

First public release of Hush Studio, an offline narration studio for Android. Write a script, pick a voice or clone your own, and the speech is generated on the phone.

Install

  1. Download HushStudio-0.1.16-arm64.apk below and open it. Android will ask you to allow installs from your browser or file manager.
  2. Open the app and go to Settings > Speech Model. Download the model (about 500 MB, one time), or import voxcpm-0.5b-q4_k.gguf if you already have it.
  3. Write a script in Create, choose a voice and tap Generate.

Requirements

  • Android 9 or later, 64-bit ARM phone
  • 4 GB RAM minimum, 6 GB recommended
  • About 650 MB free storage for the speech model

One APK works on all supported phones. Newer phones (roughly 2019 onward) use a faster engine; older 64-bit phones use a compatible engine automatically.

What's included

  • Offline speech generation with the VoxCPM 0.5B model
  • Voice cloning from 5 to 10 seconds of recorded or imported audio, with offline transcription
  • Six built-in voices from openly licensed recordings
  • Chunk studio: generate, play, edit and regenerate each paragraph separately
  • WAV export, project library, light and dark mode

Good to know

  • Generation is slower than real time. On a Moto G71 a clip takes about four times its length to generate. On older phones such as the POCO F1 (Snapdragon 845) it can take 20 to 30 times its length.
  • English only. Output is 16 kHz mono WAV.
  • Only clone voices you have permission to use.
  • Everything stays on the phone. The only network request is the model download.

Verify the download

  • SHA-256 of the APK: 628c8ab1cf5aed3fb54935b44c8b791158c3927edf686f14e10c2082cb44e96b
  • Signing certificate SHA-256: 2c36404b24fdb95fc7e75f86063e3ee4df5c0ccf67936f576ccafe869562e25a

Future releases will be signed with the same certificate.

Credits

Speech model by OpenBMB (VoxCPM, Apache-2.0), runtime by VoxCPM.cpp and ggml. Built-in voices from LJ Speech (public domain), Jenny (Dioco), Hi-Fi TTS by NVIDIA (CC BY 4.0) and the CSTR VCTK Corpus, University of Edinburgh (CC BY 4.0). Full notices are in the app under Settings > Open-Source Licenses.

Loading