Voice
Dictation, read aloud, and real-time voice conversation.
Dictation
Select the microphone in the input box and speak to transcribe your words. Check the text and send it. The send button is locked while transcription is in progress.
Read aloud
Select Read aloud under a response to have it read to you. The voice (gender and tone) and playback speed are chosen under Settings → General, in the Voice section.
Voice engine
| Engine | Description |
|---|---|
| Cloud | This is the default. |
| Local | In the desktop app, this computer processes voice. Turn on On-Device Voice Pack under Settings → Plugins and install a voice model. |
The local engine reads aloud only Korean, English, Japanese, and Chinese answers. Change the engine under Settings → General → Voice engine. If you turn off the local pack, the model remains installed. If you delete the pack, the model is also deleted.
Real-time voice conversation
This is a full-screen conversation mode where you ask questions by speaking and hear the answers. When the input box is empty, a voice conversation button is displayed in place of the send button. Voice conversation works with cloud voice.
| Item | Behavior |
|---|---|
| Response | Reading starts as soon as the first sentence is complete, without waiting for the whole answer. |
| Interruption | If you speak while Keural is talking, it stops and listens to your new question. |
| Noise | Small sounds that are not human voices are ignored to reduce unnecessary responses. |
| Camera | In the mobile app, if you turn on the camera, one frame of the screen from the moment you finish speaking is sent with your question. |
| Search | Drive search and web search are also used during the conversation. |
| Language | Korean and English. If what you say contains Hangul, Keural answers in Korean. If it is only English, Keural answers in English. |
Usage
Voice features count toward your plan's usage limit. If the limit is exceeded, a notice is displayed on the voice screen.