Audio Sources & Quality
Choose your microphone, capture computer/meeting audio, mix sources, and pick the right quality mode.
Great transcripts start with great audio. This page shows you how to pick the right microphone, capture the audio coming out of your computer (for calls, videos, and webinars), mix both together, and choose the capture quality mode that fits your room, so Kalima's transcription engine hears you clearly.
Two ways to feed audio into Kalima
Every session captures from one of two sources. You choose it in the audio source picker, available both in the session sidebar and in the setup wizard's Audio step.
- Microphone records your own voice, and anyone in the room, from any connected mic, with a live level meter and quality modes.
- Desktop / system audio transcribes a browser tab, app, or shared screen, which is ideal for calls, webinars, and videos. It is called Device audio on mobile.
You can also combine the two: when capturing desktop audio, turn on Include microphone to blend your own voice into the same transcript.
Both sources, quality modes, recording gain, and the level meter work on every plan. Picking the right combination for your situation is the single biggest thing you can do to improve accuracy.
Choosing a microphone
When Microphone is your source, Kalima records from any input device you select, such as a built-in laptop mic, a USB or headset mic, or a studio interface.
Your microphone choice is remembered, so the device you used last time is selected automatically for your next session.
Always glance at the level meter before an important session. If it stays flat while you talk, you have the wrong device selected, so switch inputs until the meter responds to your voice.
Quality modes: Enhanced vs. Raw
Microphone capture offers two quality modes. They change how your audio is processed before it reaches the transcription engine, and the right one depends on your room.
| Mode | Best for | What it does |
|---|---|---|
| Enhanced (default) | Laptop mics, shared rooms, anywhere with background sound | Turns on echo cancellation, noise suppression, and auto-gain to tame background sound and even out quiet speakers. |
| Raw | A good mic in a quiet room | Records raw, unprocessed audio, with echo cancellation, noise suppression, and auto-gain all off. Nothing is smoothed, for better or worse. |
Which one should you use?
- Stay on Enhanced unless you have a reason not to. It is the default because most sessions are a laptop mic in a room with other people, and it is the same processing your browser already applies to video calls.
- Pick Raw when you have a decent microphone and a calm setting, such as a private office or a home desk, and you would rather the audio reach the engine untouched.
Not sure which sounds better? Use the Test modes option in the audio settings to compare them on a quick sample before you commit.
Raw mode records exactly what the room gives it. Without auto-gain, a soft or distant speaker stays soft, and without echo cancellation, sound from your speakers can be picked up and transcribed as though you had said it. Use headphones with Raw mode, or stay on Enhanced.
Recording gain (input level)
If your microphone is too quiet or too loud, adjust the recording gain before you record. Gain is applied to your input level so the engine hears you at a healthy volume.
- Range: -12 dB to +12 dB, in 1 dB steps.
- Default: 0 dB (no change).
- Your setting is remembered for next time.
Open the recording gain dialog from the gear icon next to the audio source, then slide toward + for a quiet lapel mic or someone sitting far away, or toward - to trim a hot, distorting input.
Gain compensates for level, not noise. A very quiet input may still be hard to transcribe even at +12 dB, and moving closer to the mic or choosing a better device usually helps more than maxing out the slider.
Capturing computer and meeting audio
Desktop audio capture transcribes the sound coming out of your computer instead of (or alongside) your microphone. That could be a video call, a webinar, a podcast, or any video you are watching. Use the tabs below to set up the source you need.
Keep the source set to Microphone to transcribe what you and people in the room say. Select your device, watch the level meter, choose Enhanced or Raw, and record. This is the default for in-person meetings, interviews, and dictation.
How desktop audio capture works in the browser
Capturing system audio depends on your browser's screen-share support, and not every browser exposes audio in the share prompt. Some browsers only allow tab audio, not full system audio. If audio isn't coming through, use a supported browser or the Kalima desktop app, which captures system and meeting audio natively. See System & meeting audio capture troubleshooting.
Recording your screen (desktop app)
On the Kalima desktop app, when you are using Desktop audio you can also enable Record screen to save the shared screen or window as a local video file, played back in sync with your transcript.
- The video is saved to your Documents folder. It stays on your computer and is not uploaded to the cloud.
- It runs at a compact size of roughly 3 MB per minute, so even long recordings stay manageable.
- Screen recording requires the desktop app and a Plus plan or higher. On the web, on mobile, or on the Free plan, the toggle shows a lock with an upgrade or "Desktop app" prompt.
Learn more in Capturing audio & screen.
Device delay advisory
If your selected input is likely to add a delay (Bluetooth or wireless headsets, AirPods, or virtual/loopback audio cables), Kalima shows a non-blocking notice near the audio controls explaining that the delay comes from the device or environment, not the app.
This is advisory only and never blocks recording. If you want the snappiest results, switch to a wired or built-in microphone, which keeps both the audio and the live latency indicator healthier.
Wireless and virtual audio devices add latency that no app can remove. For interviews, live captioning, or any session where timing matters, a simple wired headset usually beats a premium Bluetooth one.
Best practices for clean input
- Match the mode to the room. Stay on Enhanced unless your room is quiet and your mic is good, then Raw is worth a try.
- Use the level meter to confirm the right device is active before you start.
- Sit a consistent distance from the mic and avoid bumping it.
- Prefer a wired mic over Bluetooth or virtual cables when latency matters.
- Set gain once so your voice reads at a comfortable level, then leave it.
For a deeper checklist, see Tips for accurate transcription.
Frequently asked questions
Enhanced is the default and the right pick for most rooms: it adds noise suppression, echo cancellation, and auto-gain. Switch to Raw when you have a decent microphone in a quiet room and want the audio unprocessed. You can compare both with the Test modes option before recording.
Yes. Switch the audio source to Desktop audio and, when the browser prompts you, share the right screen, window, or tab, then enable the share audio option. Turn on Include microphone if you also want your own voice captured at the same time.
Most often the browser share prompt didn't include audio, so make sure the share audio checkbox is enabled when you pick what to share. Some browsers only expose tab audio, not full system audio. Use a supported browser or the Kalima desktop app for native system and meeting audio capture.
Open the recording gain dialog from the gear icon next to the audio source and raise the level (up to +12 dB). If it is still too quiet, move closer to the mic or switch to a better input device, since gain boosts level but cannot recover audio the mic never captured.
Yes. Your microphone choice, quality mode, and recording gain are remembered, so your preferred setup is ready the next time you open the Studio.
A high-latency device, such as a Bluetooth headset or a virtual audio cable, is the usual cause. Kalima shows a device latency advisory when it detects this. Switch to a wired or built-in microphone and a stronger network connection to reduce the delay.