# Troubleshooting Fix common problems with recording, transcripts, AI providers, model downloads, imports, sharing and updates, and find out how to get help. Canonical URL: https://docs.talkbar.app/help/troubleshooting Keyboard shortcuts shown below are Windows defaults. On macOS, check the shortcut shown in TalkBar or its settings; configurable shortcuts can be changed there. Most problems have a quick fix. Start with Settings Doctor, then look for your problem below. If you're still stuck, contact the TalkBar team from inside the app. ## Start with Settings Doctor Open **Settings → Doctor** and click **Run all checks**. Settings Doctor checks your setup and points every finding at the place that fixes it. Each check shows **OK**, **Needs a look** or **Issue**, with a button that takes you to the fix. | Group | What it checks | | --- | --- | | **AI** | AI routing, your providers, Private AI, CLI tools and Ollama. **Run all checks** also sends one tiny test request to your default model. | | **Local Models** | Whether the models your features need are downloaded. | | **Voice** | Your microphone and keyboard shortcuts. | | **Notifications** | Your delivery channels. | | **Application** | Saved keys, the app version and the database. | See [Help and Doctor settings](/settings/help-and-doctor). ## Recording ### "Microphone not found" - Turn on or connect a microphone. TalkBar picks it up automatically and shows **Microphone connected**. - In **Settings → Voice → Recording**, check the selected microphone and click **Start test** under **Test audio input**. - Allow TalkBar microphone access in your system privacy settings. - Pressed `Ctrl+Shift+M`? That shortcut mutes all microphone input to TalkBar. Press it again to unmute. ### "No audio detected yet — check your microphone or audio source." - Check the audio source. **System audio** only records what your computer plays. To record yourself, choose **Microphone** or **Mic + System**. - Use the audio source menu on the recording screen to switch devices. ### "Could not start the session" If TalkBar says the voice model is not downloaded, open **Settings → Voice → Transcription** and select or download the live transcription model. Parakeet V3 is recommended. ### Speaker model warning in the New session window Large meetings need an extra speaker model. Click **Download** in the warning, or manage speaker models in **Settings → Private Local AI → System models**. ### "Recording in progress" when starting a session Another session is recording. Open it and stop it, or use **Schedule for later**. If an audio import is running, wait for it to finish or cancel it first. ### "Recording interrupted" TalkBar stopped unexpectedly, for example because the app or computer closed. Everything transcribed so far was kept. Choose **Continue** to keep recording, **Complete** to close the session and generate the summary, or **Decide later**. ### "Session audio still needs to be saved" Your transcript and summary are safe. Click **Retry saving audio** to save the audio file again. ### "Leftover recording data" on the Sessions page These recordings were cut off before a session could be created, so they can't be opened. Click **Discard** to free up space. ### "Microphone not available" in the dictation panel Dictation needs a microphone. Connect one, or allow TalkBar microphone access in your system privacy settings, then close the panel and dictate again. ### A keyboard shortcut doesn't respond Another app may already use it. TalkBar tells you when a shortcut is taken; choose a different one in Settings. **Settings → Doctor** lists any shortcuts that couldn't be registered. See [Getting around](/getting-started/getting-around#keyboard-shortcuts). ## Transcript and summary ### "Enhance transcript" isn't available **Enhance transcript** needs a completed session and its saved recording. If **Delete audio when a session ends** is on in **Settings → Voice → Recording → Saved recording**, the audio is removed after you stop, and Enhance isn't available for those sessions. ### "Summary failed" or "Summary timed out" - Click **Retry**, or **Change model** to pick a faster or more capable model. - If a local model is slow, try **Settings → Private Local AI → Find my best model**. - Very short sessions don't get an automatic summary. You can generate one from the summary panel when you're ready. ### The transcript is in the wrong language Check **Transcription language** in **Settings → Voice → Transcription**. Some speech models detect the language automatically; others transcribe only certain languages. See [Recording and transcription](/voice/recording-and-transcription). ## AI and providers ### A banner says a feature "could not use" a model - **Fix now** opens the problem in **Settings → AI Providers**. - **Use Private AI** switches to local models, if you have them. ### A provider or model doesn't work - In **Settings → AI Providers**, click **Test** next to the provider or model. - The **Advanced** tab has **Diagnostics**: provider health, route readiness and recent runtime issues. - For a cloud provider, check your API key and your account with the provider. - For a local model, check that it is downloaded in **Settings → Private Local AI**. See [Providers and models](/ai/providers-and-models) and [AI provider settings](/settings/ai-providers). ## Model downloads - **"Not enough free space"**: TalkBar shows how much space the download needs and how much is free. Free up space on the drive, then download again. - **"Download failed"**: check your internet connection, then click **Retry downloads** (during setup) or download the model again where you chose it. - Speech models are in **Settings → Voice → Transcription**. Speaker detection models and other system models are in **Settings → Private Local AI → System models**. Built-in AI models are in **Settings → Private Local AI**. - **Settings → Doctor → Local Models** shows which needed models are missing. See [Local models](/ai/local-models). ## Importing recordings ### "Unsupported file type", "The file is empty", or the file is too large or too long Use an mp3, wav, m4a, mp4, aac, flac, ogg, opus, mkv or webm file, up to 1 GB and 6 hours. ### "Audio import failed" If part of the file was transcribed, it was kept: open the session. Otherwise try again, or choose a different transcription model. See [Importing recordings](/sessions/importing-recordings). ## Exports, sharing and notifications - **An export failed.** Check that the destination folder exists and you can write to it, or reconnect Notion. **Settings → Export & Share → Export Session → Export history** has **Retry failed deliveries**. - **A share failed, or shows "Delivery unconfirmed".** Open **Manage links** or **Sharing history**, then use **Retry failed deliveries**, **Resend failed invitations** or **Send content instead**. - **Email or Telegram messages don't arrive.** In **Settings → Notifications**, click **Send Test** on the channel and fix any **Missing credentials**. - **No desktop notifications.** In **Settings → Notifications → Desktop Notifications**, click **Send Test**. **System Settings** opens the system notification settings, where TalkBar notifications may be turned off. See [Share links](/sharing/share-links), [Exports](/sharing/exports) and [Notifications](/sharing/notifications). ## Updates - **"Could not check for updates"**: check your internet connection and try again from the TalkBar logo menu (**Check for updates**) or **Settings → Check for Update**. - **"Could not download the update"**: follow the message. For example, free up disk space, then try again. - **"The update could not be verified"**: the update's signature check failed, so it wasn't installed. Try again later. - **"Could not restart to update"**: finish any recording, dictation or transcript processing, and wait for notes to finish saving, then click **Restart to update** again. See [Install and update](/getting-started/install-and-update). ## If TalkBar stops working after an update **Settings → Data & Backup → Danger zone → Reset database** starts over with an empty database and restarts TalkBar. Your sessions, transcripts, notes and knowledge are moved to a backup folder inside the TalkBar data folder, not deleted. Downloaded voice models, voice settings and speaker profiles are kept. Create a backup first if you can. See [Backup and restore](/data/backup-and-restore) before you use any reset. ## Logs Logs help the TalkBar team find the cause of a problem. They are off by default and stay on your computer. 1. Open **Settings → General → Application Logs**. 2. Turn on **Enable application logs**. For problems with built-in AI, also turn on **Enable Local AI logs**. 3. Reproduce the problem. 4. Click **Open logs folder**. Log files are in the `logs` folder inside the TalkBar data folder. You can also open the logs folder from **Settings → Data & Backup → This device**. Turn logging off again when you're done. ## Get help Contact the TalkBar team from inside the app: 1. Click **Help → Contact us** in the title bar, or **Contact us** at the bottom of the Settings sidebar. 2. Under **What do you need help with?**, pick a topic, such as **Bug report** or **General question**. 3. Enter your email, a subject and a message. You can add up to 5 screenshots. 4. Leave **Include app version and system details** on to help the team find the problem faster. 5. Click **Send message**. The team replies by email. The form doesn't attach log files. If the team asks for logs, send them by email to **contact@talkbar.app**, the address shown in the form under **Or send us a direct email:**. ## Related - [FAQ](/help/faq) - [Help and Doctor settings](/settings/help-and-doctor) - [Backup and restore](/data/backup-and-restore) - [Your first session](/getting-started/first-session)