PulseTranscription is not working
What to do, by operating system, when no audio is captured, it stops partway, or accuracy is poor.
This brings together how to narrow down symptoms such as transcription not starting, picking up only your own voice, or stopping partway. Most causes are one of three: the system's default audio device does not match the meeting tool's settings, an operating system permission has not been granted, or traffic is blocked on the corporate network. Working through it from the top gets you to the cause quickly.
Pulse transcribes by capturing your computer's microphone and system audio (the sound your computer is playing) directly in the desktop app. No recording bot joins the meeting tool, and no separate recording app is used. Two things are needed for it to work: the desktop app (macOS or Windows) running, and the note for the meeting concerned open (or a new note created). Simply sitting in the menu bar or system tray does not start recording. The macOS version requires macOS 14 or later. The web version (in a browser) only supports viewing and editing notes that already exist; it cannot record or transcribe.
When transcription starts and stops is explained in How transcription works.
Six things to try first
Before going into the detailed steps, work through these six in order. They resolve it in most cases.
Restart the app
Right-click the Pulse icon — in the Dock on macOS, in the system tray on Windows (under the “^” at the right of the taskbar) — and choose “Restart Pulse”. Where the transcription connection has been left dropped, this brings it back.
Match your audio devices to your meeting tool
The system's default microphone and speaker (on Windows, the default communication device) have to be the same as the devices selected in your meeting tool. The steps are under “Matching your system audio devices to your meeting tool”.
Check permissions
Both microphone permission and permission to capture system audio are needed. The steps are under “Checking audio permissions”.
Review Bluetooth and external devices
Where you use Bluetooth headphones or external speakers, set them as the default device (and the default communication device). Some Bluetooth and USB devices drop out; where transcription stops after a few minutes, try the built-in microphone or a different device.
Check your macOS version
Pulse requires macOS 14 or later. On anything older, updating macOS can resolve it. The version is under “Apple menu > About This Mac”.
Check the language setting
Where notes or transcripts come out in an unintended language, see Multi-language support and set the meeting language. Transcription uses the language set on the account or the workspace.
If that has not fixed it, work through the sections below in order.
Checking transcription is working
You can test this on its own, without waiting for a real meeting.
Play some audio
Play a video or some music with speech in it on your computer. Using the demo meeting video UNCHAIN, Inc. provides reproduces a situation with several people speaking.
Create a new note
Create a “New note” in Pulse.
Check system audio
A green bar moving in time with the sound at the bottom of the note means system audio is being captured.
Open the transcript panel
Clicking that bar opens the transcript panel. The audio playing appearing in grey text means it is working.
Check your own voice
Now speak yourself. Your own voice is added in green text.
Both grey text (system audio) and green text (microphone) appearing means audio capture is working. Where only one of them appears, check the settings and permissions for that device. Where neither appears, it is most likely that the app cannot reach the servers that do the transcription, so go on to the network checks.
Replace the URL of the demo meeting video with wherever it is actually published on UNCHAIN, Inc.'s site. No URL is given in this article deliberately.
Matching your system audio devices to your meeting tool
Pulse captures meeting audio using the operating system's default audio devices. Where the devices selected in a meeting tool such as Google Meet, Microsoft Teams, Zoom or Slack do not match the operating system's defaults, you get symptoms such as only the other person's voice missing, or only your own. Where the settings live depends on your operating system.
System Settings > Sound — open it and check the “Output” and “Input” tabs. On some macOS versions it is “System Preferences > Sound”.
- Output: check the device selected on the “Output” tab is the speaker or headphones actually playing the sound. Do not select a USB microphone as the output. Some USB microphones are recognised as both an input and an output device, and choosing one as the output makes it impossible to capture other participants' audio separately. Match it to the speaker selected in your meeting tool.
- Input: check the microphone selected on the “Input” tab is the same as the one selected in your meeting tool. Where you use an external microphone, select that microphone directly rather than going through a virtual audio device.
| Type | The Windows setting Pulse uses |
|---|---|
| Output (other people's voices and system audio) | Default device |
| Input (your own voice) | Default communication device |
The “System Settings > Sound” screen, with “Output” and “Input” tabs at the top and the built-in speakers selected on the output tab.
The “Settings > System > Sound” screen, showing the list of output devices with the round default-device button to the left of each.
The classic sound control panel, with a device right-clicked on the “Playback” tab and the menu open on “Set as Default Communication Device”.
Turning audio enhancements off (Windows only)
Enhancement features that come with audio devices can interfere with Pulse's audio capture. Where transcription is unstable, turn them off. macOS has no such setting.
Type “sound settings” into the Start menu, or open Settings > System > Sound.
- Click the right-facing arrow on the row for each sound device to open that device's properties.
- Find the item called “Audio Enhancements” and change it to “Off”.
- Do the same for input devices.
While you are there, try switching “Mono audio” on under Settings > System > Sound. Some audio devices do not sit well with Pulse in stereo.
The device properties screen on Windows 11, with the “Audio Enhancements” dropdown shown and “Off” available to select.
The sound device properties window, with the checkbox for disabling audio enhancements shown on the “Enhancements” tab.
The “Settings > System > Sound” screen with the “Mono audio” toggle switched on.
Checking audio permissions
Permission is needed for both system audio and the microphone. With only one of them, you end up recording only your own voice, or only the other person's. Where the settings live depends on your operating system.
- Open “System Settings”. On some macOS versions it is “System Preferences”.
- Open “Privacy & Security” and choose “Microphone”.
- Find Pulse in the list and switch it on.
- Switch Pulse on under “Screen & System Audio Recording” in the same “Privacy & Security” section too. On some macOS versions the item is shown simply as “Screen Recording”.
Immediately after changing a permission, you may be asked to restart the app. If you are, restart it and test again.
Once permissions are set, run the test under “Checking transcription is working” again. Both grey and green text appearing means the permissions are granted correctly.
Antivirus and endpoint protection products (Windows)
Where permissions and audio devices are both correct and audio still is not captured, check whether a security product on the machine is blocking access to audio devices.
Kaspersky
Kaspersky antivirus products are known to block Pulse's access to audio devices, stopping transcription working. Where you have Kaspersky and transcription does not work, add Pulse to Kaspersky's allowlist. The steps are in Kaspersky's help (allowing an application to access audio devices).
Other endpoint protection products
At Japanese companies, endpoint protection products distributed to machines (EDR, asset management and data-loss prevention software of various kinds) sometimes restrict applications' access to the microphone or system audio, or their traffic to the outside. Check the following with your IT department.
- Whether per-application control of audio device use is enabled
- Whether Pulse's executable is being monitored or blocked as an unknown application
- Whether Pulse's destinations appear in the traffic blocking logs
Some products return no error to the app when they block: the audio is simply treated as silence. “Permissions are right but the green bar does not move at all” is the classic sign of this kind of blocking.
Reviewing audio mixing hardware and software
Where you use hardware or software that adds extra processing to the audio — a USB microphone amplifier, an audio interface, mixing software that combines several microphone inputs — it can interfere with Pulse's capture. If none of that applies to you, skip this section.
If disabling or bypassing them temporarily makes Pulse work, that is where the cause lies. To carry on using them, check the following.
- Turn pass-through (loopback) off — this is the setting that sends microphone input to both the input and the output. Left on, your own microphone audio is transcribed as another participant's voice.
- In stereo, send the audio to the left channel — Pulse assumes mono audio. Where the device is set to stereo it reads the left channel, so make sure audio is sent to the left (or to both).
- Make sure audio reaches both your call software and Pulse — where the voice chat or microphone path goes only to the meeting tool, no sound reaches Pulse.
Virtual desktop environments (VDI / Citrix / Azure Virtual Desktop)
Where you run Pulse inside a virtual desktop session, only microphone audio may be captured and other participants' audio (system audio) may not. In that case only what you say is transcribed and the other side is missing.
The remedy is to install and run Pulse on the machine in front of you rather than inside the virtual desktop session. Where your internal rules require it to be used on the virtual desktop, you will need to discuss an audio redirection configuration with your IT department.
Audio quality drops on a Bluetooth headset
Where quality drops the moment transcription starts, that is a constraint of how Bluetooth headsets work. Most headsets cannot handle high-quality audio output and microphone input at the same time. When Pulse uses the headset's microphone, the headset switches to the lower-quality call mode.
To avoid this, set your computer's built-in microphone as the default input device in the system sound settings. The headset is then output only, and keeps its quality.
On macOS, the default input device setting can revert every time a Bluetooth headset is disconnected and reconnected. Set it again after each reconnection.
Restarting your computer
Where the process used to capture system audio is left stopped, restarting the app does not bring it back but restarting the computer can. If the steps so far have not resolved it, restart the whole machine.
Where the network is the cause
Where you have been through audio devices, permissions and restarting and transcription still does not work, the network, a VPN, a proxy, a firewall or a web filtering product may be the cause. Transcription uses a long-lived secure WebSocket connection, which makes it particularly sensitive to how a corporate network is configured. Where the start of a meeting is transcribed but it stops after a few minutes, a VPN or proxy connection-lifetime setting may be involved.
The steps for narrowing it down, the symptoms blocking produces, the list of destinations to allow and a template request for your IT department are all in Network and allowlists. Where you are on a company-issued machine and cannot change operating system permissions or audio device settings yourself, that same template has a section for permissions.
If none of this resolves it
If the steps above do not resolve it, contact tharada@the-unchain.com. Including the following helps us investigate faster.
- What the symptom is (nothing transcribed at all / only your voice / only the other person's / stops partway)
- Where it stops partway, roughly how long it runs before stopping
- The audio device you were using (built-in microphone, Bluetooth headset, USB microphone and so on, including the model)
- The meeting tool you were using (Zoom, Google Meet, Microsoft Teams and so on)
- Your operating system and version (for example macOS 15.3, Windows 11 23H2) and the Pulse app version
- The result of the test under “Checking transcription is working” (whether grey text and green text appeared)
- Your network environment (corporate LAN, VPN or not, proxy or not)
On a company-issued machine, where you do not know how the endpoint protection product or proxy is configured, getting in touch together with someone from your IT department reduces the back-and-forth.