Loquela

Beta

Loquela DesktopThe voice studio for your computer.

Dictate into any app, transcribe files, clean up noisy audio and run speech models on your own computer. Pick the model, patch it for your field and change every setting.

Loquela Desktop's Home page with the dictation key and a waveform
The dictation indicator showing hands-free dictation on this computer

Dictation comes first

Hold your key and talk, or tap it once and keep talking hands-free. The words land wherever your cursor is, in any app, and a small indicator by the cursor shows what's happening.

  • Any key, chord or mouse button as the dictation key
  • Paste the last dictation again, or undo it in one step
  • A key per language, if you speak more than one
  • Music turns down or pauses while you talk

Talk. It types. Tune it.

  1. Hold your key and talk

    Right Ctrl to start, or any key you choose.

    Right CtrlThe dictation indicator: a waveform, the time and the word Local
  2. The words land in your app

    Notes, email, chat or code: wherever the cursor is.

    Notes

    Send the updated contract to Dana by Thursday and book the small room

    The dictation indicator: a waveform, the time and the word Local
  3. Make it yours

    Switch models, add a patch pack, change any setting.

    A patch pack switched on for a model, with its strength slider
The Models page in Loquela Desktop, with the verified catalog and a Hugging Face tab

Your model, every setting

Setup measures your computer and suggests a model that fits; after that the choice is yours. Browse the verified catalog or search Hugging Face, keep several models, and open every engine setting when you want them.

  • Simple, Advanced and Developer levels, so the controls never get in the way
  • A profile per app: its own model, words and formatting
  • Your vocabulary, and corrections it learns from when you allow it
  • Bring your words and settings over from Wispr Flow, Superwhisper or Handy

Model patching

Patch packs teach a speech model the words of a field, like medicine, law or code. Stack several, set how strongly each one applies, and switch them per model. Loquela makes the first packs, and you can import pack files.

The Patch packs page: a coding pack on at 60% strength, more packs to download

A whole voice studio

Dictation is the start. The same app works on recordings and audio too.

Transcribe files

Drop in recordings and get text, with speaker labels.

Denoise

Clean up noisy audio before it's transcribed.

Translate

Speak one language and get text in another.

Read aloud

A key that reads the selected text back to you.

Plug it into your other tools

Turn on the local server and other apps on your computer can use Loquela as their speech engine, through an OpenAI-compatible API. It's off until you switch it on, answers only this computer unless you allow your network, and never stores what it transcribes.

Bring a cloud when you want one

Everything above runs on your computer. When a machine can't run the models, is already busy with your own AI work, or your volume makes hosted transcription the better deal, connect Loquela Remote with an API key, with PGP encryption on Pro and Business plans, or use your own key from OpenAI, Groq or Mistral.

Your words, as you said them

No rewriting

Loquela doesn't run your dictation through a language model to polish it. What you said is what you get.

Offline lock

One switch stops every connection the app makes, model downloads included.

Locked history

Your dictation history is encrypted on your computer and opens with Windows Hello.

Price: free

Free for personal use, and for work at organizations with up to 10 people and under US$250,000 a year. Bigger organizations, and anyone serving Loquela Desktop to others for money, need a licence.

Try the beta

Loquela Desktop is in beta: it works, and it changes often. Tell us what you'd change.