New in Breeze: voice notes for tasks, comments, notes, and chat

We added voice notes to Breeze so you can record short audio updates directly inside your work. Voice notes can be used on tasks, comments, notes, and chat messages.

You can also transcribe recordings to text in real time. This means you can speak naturally, see the text appear in the editor, and keep the original audio clip attached when needed.

Breeze task editor showing real-time voice note transcription with an audio clip

Where voice notes work

Voice notes are available in the Breeze editor. You can use them in the places where project communication already happens:

  • Tasks
  • Comments
  • Notes
  • Chat messages

The recording is added directly to the item you are working on. A voice note added to a task stays with that task. A voice note added to a comment, note, or chat message stays there too.

How to record a voice note

Open the editor and click the new microphone icon in the toolbar.

After you click it, the recording controls appear at the bottom of the screen. You can start recording, stop recording, and choose what Breeze should insert into the editor.

Breeze task editor with voice note recording available width=

If transcription is enabled, Breeze inserts the transcribed text into the editor while you are recording. When the recording ends, Breeze can also insert the audio clip into the editor.

You can change these options from the dropdown menu in the recording controls. Transcription and audio clip insertion can be turned on or off separately. This gives you three simple options:

  • Insert only the transcript
  • Insert only the audio clip
  • Insert both the transcript and the audio clip

Why voice notes are useful

Voice notes are helpful when typing is slower than speaking, or when a written message would lose too much context.

For example, you can use a voice note to explain a task, describe a bug, give feedback on design work, record a quick project update, or capture an idea before it gets lost.

The transcript makes the update easy to read, edit, scan, and search. The audio clip keeps the original recording available for anyone who wants the full context, tone, or detail behind the message.

This is especially useful for teams that work across different schedules. One person can leave a clear spoken update, and the rest of the team can read the transcript or listen to the recording later.

Supported languages

Voice note transcription supports English, Chinese, Hindi, Spanish, Arabic, French, Portuguese, Russian, German, Japanese, Korean, Italian, and Dutch.

This makes it useful for international teams, multilingual clients, and projects where people do not always work in the same language.

Audio and transcription handling

A voice note has two parts: the original audio clip and the optional text transcript.

The pure audio clip stays in Breeze. When you add a voice note to a task, comment, note, or chat message, the audio file is stored with that item in Breeze.

Breeze voice note options for inserting a transcript, an audio clip, or both

If transcription is enabled, the audio is processed by AI in the cloud using a third-party transcription provider. The audio is sent for the purpose of creating the transcript, and the resulting text is inserted back into Breeze.

This means the audio clip remains stored in Breeze, while AI transcription is handled in the cloud by a third party.

Technical implementation

Breeze uses Voxtral, an audio model family from Mistral AI, for voice note transcription.

Voxtral is designed for speech understanding and audio transcription. It can process spoken audio and turn it into text, which makes it suitable for short voice notes inside project communication.

In Breeze, transcription runs while the user is recording. The spoken words are converted into text and inserted into the editor in real time. This keeps the workflow close to writing a normal comment or note, but without requiring the user to type everything manually.

Mistral also provides Voxtral Mini Transcribe V2, a transcription-focused model that supports multilingual transcription, timestamps, longer audio processing, and speaker diarization.

Using Voxtral lets Breeze add real-time transcription without building a speech recognition system from scratch. The result is a simple editor feature for users, backed by a dedicated audio transcription model in the background.

A small update that makes communication easier

Voice notes are meant for the small moments where speaking is faster than typing. A task explanation, a quick update, a client comment, or a note to yourself can now be recorded directly where the work is happening.

With real-time transcription, the message is still easy to read, edit, and search. With the audio clip, the original context is still there when people need it.

Voice notes are now available in the Breeze editor for tasks, comments, notes, and chat messages.