dGENIX
Log inJoin waitlist

Voice Chat

You can talk to your assistant instead of typing. dGENIX has two voice modes: Voice Memo (quick dictation) and Voice Call (a fully spoken conversation).


Voice Memo, quick dictation

When to use it? When you want to send a question or command quickly without typing. Your assistant replies with text in the chat.

How it works:

  1. Click the microphone icon to the right of the text field in the chat
  2. Speak your question or command
  3. Click again to stop
  4. Your question is converted to text and processed
  5. The answer appears as text in the chat

Credit cost: transcription + AI processing, roughly 25-200 cr depending on the command.


Voice Call, a fully spoken conversation

When to use it? When you really want a conversation with your assistant, ask, listen, respond, without typing. The assistant speaks its answers aloud.

How it works:

  1. Click the sound-waves icon to the right of the text field
  2. An animated orb screen opens, this is the Voice Call interface
  3. Speak your question or command
  4. The assistant processes and speaks the answer back
  5. After the answer it listens again straight away, no clicking needed
  6. Click "Hang up" to end the call

The orb: the animated sphere reacts to sound, it pulses when you speak, settles when you are quiet and moves while the assistant is thinking.

Text shown in parallel: while the assistant speaks, its answer also appears as text on screen.

Voice Call credit cost:

Action Credits
Transcribe your question ~5-20 cr
AI processing (Haiku) ~25-50 cr
Spoken answer ~500 chars ~1,375 cr
Spoken answer ~1,000 chars ~2,750 cr

Voice Call uses more credits than typing, mainly because of text-to-speech.


Setting a voice

You choose which voice your assistant uses for Voice Chat. Go to Settings -> AI Assistant -> Voice.

There are two groups:

  • Dutch voices, for use on app.dgenix.nl
  • English voices, for use on app.dgenix.com

The voices are powered by ElevenLabs, one of the most realistic text-to-speech systems available.


Voice messages via Telegram

If you have the Telegram add-on, you can also send voice messages via Telegram:

  1. Send a voice message to @dGENIXbot
  2. The assistant transcribes it automatically
  3. It processes the command and replies as text in Telegram

Telegram voice messages do not return a spoken answer (that is only in Voice Call via the dashboard).


Tips

For the best transcription:

  • Speak clearly and not too fast
  • Be in a quiet environment
  • Finish your sentence before you stop

Voice Memo or Voice Call?

  • Use Voice Memo for short commands ("create a calendar appointment", "write an email")
  • Use Voice Call for a full conversation, brainstorming or discussion

Be credit-aware: Voice Call is the most expensive channel. It is most enjoyable for longer conversations. For short commands, typing or Voice Memo is more efficient.


What you see

You open a Voice Call and ask out loud what is on today. The orb pulses while you speak, settles when you stop, and moves while GENI is thinking. Then you hear the answer and read along:

You have three appointments today. At ten you are calling De Vries about the quote, at half past one you are at the supplier, and at four you have a team meeting. Between twelve and half past one is your only free block.

Shall I prepare the De Vries call?

You do not have to click anything: after the answer it listens again straight away. Say nothing more and it waits quietly until you hang up.

Limits

  • Voice Call is the most expensive channel. Text-to-speech costs the most credits; for short requests typing or a Voice Memo is cheaper.
  • Telegram does not speak back. There you get text; only the Voice Call in the dashboard talks.
  • Background noise costs you accuracy. In a busy room or on speaker it degrades noticeably.
  • There is no live interruption. You cannot override GENI mid-answer; wait until it finishes or hang up.
  • The voice follows the domain. Dutch voices on app.dgenix.nl, English on app.dgenix.com.
  • Confirmations do not work by voice. If you ask for something irreversible, GENI points you to the chat, because that is where the button is.

Troubleshooting

You are not being heard. Grant your browser microphone permission and check that the right microphone is selected in your system settings.

The transcription is wrong. Speak more slowly and finish your sentence before stopping. Jargon and proper names go wrong more often; type those instead.

You hear no answer. Check your volume and whether the tab is muted. If it stays silent while the text does appear, the speech conversion failed; try again.

It uses credits faster than expected. That is correct: a spoken answer of a thousand characters costs about 2,750 credits. For short requests Voice Memo is considerably cheaper.

The wrong language comes out. The voice follows the domain you work on. Use app.dgenix.com for English.

Frequently asked questions

What is the difference between Voice Memo and Voice Call? A memo is recording once and getting a text answer back. A call is a running conversation where GENI answers out loud and then listens again.

Can GENI do everything it does in chat? Yes, the same skills and the same memory. Only confirmations for irreversible actions go through the chat.

Is my voice stored? The recording is used to turn your words into text; we keep the conversation as text, like any chat.

Does it work on my phone? Yes, in the browser. Make sure you allow the microphone.


Next: AI assistant · The credit system · Settings On the site: all skills · pricing

dGENIX Docs, Voice Chat, updated August 2026

Keep reading on the blog

Get started