Deepgram & ElevenLabs

Deepgram and ElevenLabs can turn your meeting audio or dictation into text in Kimiko. You need an account with your chosen provider, an API key, and an internet connection. No coding is needed.

Cloud speech is optional. You can keep summaries and chat on local AI while using either service for transcription.

Free credit to get started

These offers were checked on 28 September 2026:

Provider	Free allowance
Deepgram	US$200 in signup credit for new accounts, with no credit card required. This is a starting balance, not a monthly allowance.
ElevenLabs API	Its Free / Pay as you go column lists 2 hours 30 minutes of Scribe v2 Realtime transcription.

ElevenLabs also advertises 10,000 free credits per month on its ElevenCreative plan. Kimiko uses the API, whose current pricing is in dollars and audio time. Check your API allowance rather than treating creative credits as recording minutes.

Offers can change, and the provider’s audio allowance is not a guarantee of the same duration of Kimiko meetings. Check your dashboard for your remaining balance before a long recording. Any provider charges are separate from your Kimiko subscription.

Get a Deepgram key

1. Create an account or sign in to the Deepgram Console. Select your project; new accounts have one created automatically.
2. Open Settings → API Keys → Create a New API Key. Give it a name such as Kimiko, choose Member permissions, and set an expiry that suits your use.
3. Create the key and copy its secret. Deepgram only shows the full key once. Continue with Connect it in Kimiko below.

See Deepgram’s key creation guide for the current console steps.

Get an ElevenLabs key

1. Sign in to ElevenLabs, then open Developers → API Keys if needed.
2. Create a key named Kimiko. Keep it restricted and enable Speech to Text access. You can also set a credit limit for this key.
3. Copy the full key when it appears; it will be hidden afterward. Continue with the steps below.

See ElevenLabs’ API key guide for its current controls.

Before sending audio, review ElevenLabs’ data-use settings. To opt out of model training on new submissions, open your profile menu → Terms and privacy → Data use, turn off Improve the models for everyone, and save your choice.

Connect it in Kimiko

1. Open Settings → Transcription → Speech models → Cloud.
2. Click the key icon beside Deepgram or ElevenLabs, paste the key, and choose Save.
3. Open the key icon again and choose Test. Resolve any error before continuing.
4. Choose Use beside Nova 3 (live) for Deepgram or Scribe v2 Realtime for ElevenLabs. Review the audio-sharing prompt and choose Use cloud transcription to confirm.
5. Check Meeting language, then record a short sample and check the transcript. The selected provider also transcribes imported recordings.

If you see Blocked (local-only), cloud access is disabled. To use it, turn off Local-only mode in Settings → Privacy. Meeting audio, including other participants’ voices, will be sent to the provider you select.

Use it for dictation too

Open Settings → Dictation → Speech model, select the cloud model, and confirm Use cloud dictation when prompted. Dictation has its own model and language settings; selecting a meeting provider does not change them automatically. See the dictation guide for shortcuts and formatting.

If it does not connect

• Key rejected: check that you copied the full secret, that it has not expired or been revoked, and that an ElevenLabs key permits Speech to Text.
• Test passes, but transcription fails: check the provider’s balance, model access, and usage limits. The connection test does not transcribe audio or guarantee remaining credit.
• Cloud access blocked: check your internet connection and Kimiko’s Local-only mode.

Keep keys private and enter them only in the desktop app. To return to local processing, select a downloaded model in Transcription and review Dictation separately. See cloud providers for AI settings and privacy and data for more on what leaves your device.