# Your first session Set up, record and review your first TalkBar session from start to finish, or create one from an existing audio file. Canonical URL: https://docs.talkbar.app/getting-started/first-session This walkthrough takes you from an empty workspace to a finished session with a transcript and summary. It takes a few minutes. If you haven't finished [first-run setup](/getting-started/first-run-setup) yet, do that first: it downloads the speech and speaker models every session needs. ## Before you start - **Microphone.** Connect the microphone you want to use. In **Settings → Voice → Recording** you can pick it and click **Start test** under **Test audio input** to hear how it sounds. - **Microphone access.** Allow TalkBar to use the microphone in your system privacy settings. - **Speech model.** Transcription runs on this computer. If the model is missing when you start, TalkBar tells you to download it in **Settings → Voice**. - **AI.** Summaries and titles need an AI model. You chose one during setup; to change it, click the TalkBar logo and choose **Set up AI…**. ## 1. Set up the session Click **New session** in the title bar or on Home. The New session window opens. 1. **Template.** Pick the kind of session, such as *Meeting*, *1:1* or *Voice Note*. The template decides the summary sections. **More…** shows all templates. 2. **Audio.** Choose what to record: - **Microphone** for your own voice, such as voice notes or an in-person meeting; - **System audio** for the other side of a call, played by your computer; - **Mic + System** for both sides of a call. This is the usual choice for online meetings. 3. **Speakers.** Roughly how many people will talk: **Just you**, **You + 1**, **Small (2–4)** or **Large (5–8)**. This helps TalkBar tell voices apart. For large meetings TalkBar may ask you to download an extra speaker model. 4. **Optional.** Choose the AI model for the summary, add topics, file the session under a project or series, attach knowledge cards and documents, and adjust **AI & privacy**. 5. Click **Start session**. Click **Save as my defaults** before starting to keep these choices for next time. Not ready yet? **Schedule for later** sets a date and time instead. If **Start session** doesn't work, check for a **Recording in progress** notice (another session is recording; stop it first), an audio import that is still running, or a missing audio device. ## 2. Record - The **Recording — let everyone know** reminder appears. Telling everyone present that they are being recorded, and getting their consent where the law requires it, is your responsibility. Click **Got it**. - The screen shows **Getting audio ready…**, then **Waiting for someone to speak** until the transcript starts. If it says **No audio detected yet — check your microphone or audio source.**, check that the right microphone and audio source are selected. - The live transcript appears by speaker. Click a speaker chip to rename the speaker, so the transcript and summary use real names. - The timeline sums up the conversation as it goes. Use **Ask AI** to ask about what has been said so far. - **Pause** and **Stop** are in the recording controls at the top. See [Recording a session](/sessions/recording) for everything on this screen. ## 3. After you stop - **Very short sessions.** If almost nothing was recorded, TalkBar asks **Keep this short session?** Choose **Keep session** or **Move to Trash**. If you don't choose within about 15 seconds, the session moves to Trash. **Undo** in the notification brings it back. - **Summary.** For a normal session, TalkBar writes the final summary automatically and suggests a title. - **Continue or complete.** The session is now ended but still open. **Continue session** records more into the same session. **Complete session** locks it and opens its details page. You can record a linked follow-up afterwards. ## 4. Review the result Open the session from **Sessions** (or from the sidebar, where recent sessions are listed). The session details page shows: - the **summary**, organized into the sections of the template. Edit any part, or regenerate it in another language, length or model; - the **transcript**, **action items**, **notes**, **documents** and **knowledge** for the session; - **Ask AI**, to ask questions about the session. Answers show the sources they are based on; - **Enhance transcript**, which re-processes the recording on this computer with a more accurate model to improve the text and who said what; - **Share**, to publish a read-only link or send the content by email or Telegram, and the session menu, to export it. See [Session details](/sessions/session-details), [Share links](/sharing/share-links) and [Exports](/sharing/exports). ## Or start from an existing recording You can create a session from an audio or video file instead of recording live. 1. Click **New session** and choose **Upload audio file**. 2. Drop a file, or choose one. Supported formats are mp3, wav, m4a, mp4, aac, flac, ogg, opus, mkv and webm, up to 1 GB and 6 hours. 3. Pick the transcription and summary models, then click **Start import**. A progress window shows each stage. If you keep it open, the session opens when it's done. Click **Process in background** to keep working; a notification says **Audio import complete** when it's done, and **View session** opens it. See [Importing recordings](/sessions/importing-recordings). ## If something goes wrong Open **Settings → Doctor** and click **Run all checks**, or see [Troubleshooting](/help/troubleshooting). ## Related - [Starting a session](/sessions/starting-a-session) - [Speakers](/voice/speakers) - [Scheduling and preparing](/sessions/scheduling-and-preparing) - [Templates and skills](/sessions/templates-and-skills)