Introduction
TalkBar records your meetings, calls and conversations right from your desktop, turns them into a transcript and summary, and remembers what matters. It is local-first: your sessions, notes and knowledge are stored on your computer, and you decide whether AI runs on this computer or with a provider you connect.
Bot-free capture
TalkBar records from your own computer. It uses your microphone for your voice and your computer's audio for the other side of the call. No bot joins the meeting and nothing is added to the call.
Because it captures your computer's audio, it works with Zoom, Microsoft Teams, Google Meet, other meeting apps and calls in your browser. You choose what to record for each session: Microphone, System audio, or Mic + System.
Local-first privacy at a glance
- Transcription runs on your computer. Speech-to-text uses a speech model downloaded to this device.
- AI runs where you choose. Use Built-in AI (a model downloaded to this computer), Ollama, an AI tool installed on this computer, or a cloud provider with your own API key. Each AI feature can use its own model.
- Private AI Mode keeps AI on local models only. One switch blocks cloud and CLI providers.
- Your data stays in one folder on your computer. Open it from Settings → Data & Backup → Open data folder. You can open it, back it up to a ZIP file and restore it.
Apart from what you start or set up yourself, such as exports, share links, notifications and requests to your own AI provider, TalkBar uses the network for model downloads, update checks and anonymous usage statistics. You can turn usage statistics off in Settings → General → Share usage statistics. Recordings, transcripts, notes and chats are never included.
Read more in How TalkBar uses AI, Private AI Mode and Privacy and your data.
Before, during and after a conversation
| When | What you can do |
|---|---|
| Before | Schedule a session with reminders, and prepare it with notes, documents, knowledge cards and open action items from last time. |
| During | Follow the live transcript by speaker, a running summary and timeline, knowledge cards that appear when the conversation reaches their topic, and Ask AI about what has been said so far. |
| After | Review the summary, decisions and action items, ask questions about the session, share a link or export it, and find it later with search. |
Between conversations you can also write notes, chat with your sessions and documents in Chats, dictate into any app, and keep the facts worth remembering as knowledge cards.
System requirements
| Requirement | Details |
|---|---|
| Operating system | Windows or macOS. Check the release for supported versions and hardware. |
| Disk space | Room for the speech, speaker and AI models you download, plus your recordings. First-run setup shows the size of each download and the free space on the drive. |
| Internet | Needed to download the installer and models, and for cloud AI providers, sharing and updates. Transcription and Built-in AI run on this computer once their models are downloaded. |
| Audio | A microphone to record your own voice. Without one you can still record system audio. |
Local AI models run faster on more capable computers. During setup, the Model Advisor measures this computer and marks the model it recommends.