Support

The iOS release is being prepared.

Getting started

1. Connect to Wi-Fi, leave at least approximately 1 GB free, and download the model.
2. While it downloads, use Register voice to edit a script and record, or choose File to import audio of your own voice. Recording requires microphone permission.
3. When the model is ready, choose a speech language in Create and enter a short sentence.
4. Keep the app in the foreground during generation. Play, save or share the result from History.

Import an audio file

Open Register voice → File and select WAV, M4A or MP3 audio. Files must be 3–15 seconds and no larger than 20 MB. Trim longer audio before importing. You can optionally enter the words spoken in the file. The original file is unchanged. Cloud-hosted files may need a network connection.

Playback volume

Quiet generated audio and previews receive up to approximately 6 dB of gain with peak headroom. ORVIO does not change your device volume. Start at a low volume and adjust for your speaker or headphones.

Generation is slow

Generation runs locally on the CPU and may take minutes depending on the device and text. Start with one short sentence. Let a hot device cool down. iOS limits each generation to 80 characters and approximately 11 seconds; output may stop mid-sentence at the limit.

Pronunciation or voice quality

Record naturally in a quiet place at a consistent microphone distance, avoiding background music, echo and clipping. Match the speech language to your text. Voice similarity and accent vary by recording and language. Language selection does not translate text.

Resume a download

Check storage and connectivity, then resume in Settings. Removing the model does not remove voices or history. Do not casually uninstall to troubleshoot: uninstalling also removes local audio.

Microphone access

Check ORVIO’s microphone permission in iPhone Settings. Recordings stop after 15 seconds and are interrupted when you leave the app.

Appearance and language

Select System, Light or Dark in Settings. The interface supports Japanese and English. Speech language is a separate setting.

Delete your data

Use the menus in Register voice and History. Remove models in Settings. Delete exported or shared copies separately at their destinations.

For bug reports, include your device, OS and app version, the steps taken, and the error message. Japanese and English inquiries are welcome.