Audio, recording, and subtitles

Sound and captions round out a video: balance your tracks, shape a voice or sound effect, remove noise, capture new footage, and caption the result.

Quick overview

1. Open the audio mixer to see a fader for every track plus the master bus.
2. Set volume, pan, mute, and solo per track while you listen.
3. Apply a voice or audio preset to a clip for an instant tone change.
4. Clean up background noise with one-click AI denoise.
5. Record your screen, camera, or microphone directly onto a track.
6. Generate subtitles automatically, or import an existing file.

Detailed reference

Mixer

Control	Function
Volume fader	Per-track output level, 0 to 200%
Pan	Position a track in the stereo field
Mute	Silence a track without removing it
Solo	Hear only this track
Master bus	A single fader for the combined mix
Meter	A live level meter with peak hold on every strip

Open the mixer from Audio Mixer in the toolbar's Tools menu. Only individual track strips have pan, mute, and solo; the master strip carries just the combined volume fader and meter.

Clip volume and EQ

Each clip also has its own Volume slider and a simple three-band EQ (Low, Mid, High, each up to ±12 dB) in the inspector, for quick per-clip tone shaping without opening the mixer.

Voice and audio presets

Apply a preset to reshape a clip's sound in one click. Each preset combines an equalizer curve with compression, reverb, or another effect tuned for the sound.

Preset	Character
Telephone	Narrow bandpass, classic phone-call tone
Radio	Vintage AM radio
Echo	Hall reverb with delay
Robot	Ring modulation, metallic
Underwater	Deep low-pass, muffled
Bass Boost	Warm, boosted low end
Treble Boost	Extra air and presence
Vocal Clarity	Presence boost with de-essing
Cinematic	Cathedral reverb, warm
Lo-Fi	Vinyl-style warmth
Megaphone	Midrange peak, gritty
Whisper	Soft, breathy, airy

An Intensity slider (0 to 100%) scales how strongly the preset's equalizer, compression, and reverb are applied, and Reset removes the active preset entirely. Applying a preset acts on every clip you currently have selected, not just one, so you can restyle several clips in a single click.

AI denoise

One-click noise removal for a clip's audio, powered by an on-device neural model (RNNoise). It runs entirely in your browser, with a wet/dry mix slider to blend the cleaned signal back with the original.

Recording

Capture new footage without leaving the app: open the media/add panel (the same place you add a file or browse the gallery) and choose one of four capture modes.

Mode	Captures
Audio	Microphone only
Camera	Webcam and microphone
Screen	Your screen, with microphone audio mixed in
Screen + Camera	Your screen with a circular webcam picture-in-picture in the corner

Recordings always save to your media library. If the video editor is open and active when you stop recording, the clip also lands straight on a track, ready to edit; if it isn't, pull the recording in afterward from the media browser (Alt + 8).

Auto-subtitles

Generate timed subtitles from project audio with one of two methods:

Method	Provider	Result
Automatic	ElevenLabs Scribe	Cloud transcription with multilingual support
Automatic	Deepgram Nova-3	Cloud transcription with word timing and optional speaker separation
Local	Whisper Turbo	On-device transcription through WebGPU

Choose Automatic to reveal the cloud provider selector. Choose Local to keep audio on the device. Local Whisper downloads once and is cached in the browser. Deepgram speaker separation stores anonymous speaker labels and timing with the subtitle cues so speaker-aware dubbing can reuse who spoke when. Provider and model choices follow AI Settings; switching one off removes it from the editor.

When Deepgram finds multiple speakers, CardCraft creates one color-coded subtitle track per speaker. Tracks use stable names such as English Speaker 1 and English Speaker 2. Their colors, speaker identities, cue timing, translation lineage, preview layout, and export layout survive save and reload. Simultaneous speakers remain on separate visual lanes instead of covering one another.

Subtitle Translation uses the current text-model catalog from AI Settings. It processes long timelines as durable batches of at most 40 cues and 8,000 source characters. Completed batches appear on the canvas immediately. You can pause, resume, or cancel without losing finished work, and a source-track revision guard prevents an old translation from overwriting captions edited after the job started.

AI voiceover and dubbing

Voiceover offers ElevenLabs and Deepgram Aura-2 as independent providers. The available voices come from AI Settings. Deepgram Aura voices are currently published for English, Spanish, Dutch, French, German, Italian, and Japanese. Unsupported target languages are blocked before generation.

Dubbing offers two workflows:

Workflow	How it works
Automatic ElevenLabs	Existing one-click dubbing translates the selected clip and preserves voices through ElevenLabs
Deepgram Cue Dubbing	Uses existing subtitle cues, assigns one Aura voice to each speaker, and generates every cue directly through Deepgram on a separate speaker audio track

Deepgram Cue Dubbing does not require ElevenLabs. If the selected subtitle set already contains speaker metadata, it does not transcribe the media again. If speaker metadata is missing, CardCraft can analyze speakers once with Deepgram Nova before generating speech.

Cue dubbing keeps original start times. CardCraft measures each generated audio segment, uses Deepgram speed control first where the selected language supports it, and then applies bounded playback fitting. Lines that still exceed their source window are marked for review. CardCraft never cuts or retimes video automatically. Completed cue assets are cached and checkpointed, so reopening or retrying the workflow reuses finished audio instead of paying for it again.

Captions: import and export

Bring in an existing .srt or .vtt file if you already have captions: adding a subtitle track from the track list opens a file picker immediately, so cancel it if you want an empty subtitle track instead (auto-subtitles will create one for you automatically when you run it). After auto-subtitles finishes transcribing, download that batch of cues as SRT or VTT directly from the same dialog. A subtitle track renders directly into the video frame whenever it is not muted, so mute the track if you want to export without captions.

Step by step

1. Balance a multi-track mix

1. Open Audio Mixer from the toolbar's Tools menu; you'll see a strip for every track plus the master bus.
2. Play back your timeline and watch the meters; drag a track's fader to bring it up or down, and use Pan to place it in the stereo field.
3. Mute a track to check the mix without it, or Solo one to hear it in isolation while you set its level.
4. Use the master fader for the overall output level once the individual tracks feel balanced.

2. Reshape a voice with a preset

1. Select the clip (or clips) you want to restyle.
2. Open Audio FX from the Tools menu and click a preset card, Bass Boost and Telephone are good starting points for a quick, reliable change.
3. Drag the Intensity slider to taste, or click Reset to remove the preset and start over.
4. If a preset with reverb (Echo, Robot, Cinematic, Lo-Fi, Whisper) doesn't seem to change the sound, try a non-reverb preset instead (Telephone, Underwater, Bass Boost, Treble Boost, Megaphone) while that gets sorted out.

3. Record a voiceover and drop it on a track

1. Open the media/add panel and choose Audio to record your microphone alone, or Camera / Screen / Screen + Camera for video capture.
2. Speak your voiceover, then stop the recording.
3. With the video editor open, the recording lands directly on a track at the playhead; otherwise, open the media browser (Alt + 8) and drag it in from your library.

4. Auto-generate and download subtitles

1. Select the clip with the dialogue you want captioned.
2. Run auto-subtitles, choose Automatic or Local, select the provider when Automatic is active, then choose a language or leave it on auto-detect.
3. Review the generated cues on the subtitle track, then download them as SRT or VTT from the same dialog if you need the file for another platform.

Common tasks

Task	Steps
Check a mix without background music	Click Mute on the music track in the mixer instead of deleting or moving the clip
Apply a voice preset to several clips at once	Select all of them first (Shift-click or drag-select), then click a preset card; it applies to every selected clip
Silence a video clip's built-in audio	Set its Volume to 0 in the inspector, or detach the audio to its own clip first if you want to keep it elsewhere
Export without captions	Mute the subtitle track before exporting; it renders into the frame whenever it isn't muted

Troubleshooting

• Denoise doesn't seem to reduce noise. Enabling it updates the clip's setting and shows a confirmation, but it may not currently produce an audible change. If a clip is still noisy after enabling denoise, don't assume you're missing a step, try trimming the noisy section or replacing the audio instead.
• A voice preset with reverb doesn't seem to apply. Echo, Robot, Cinematic, Lo-Fi, and Whisper can fail to take effect. Telephone, Underwater, Bass Boost, Treble Boost, and Megaphone are the more reliable picks if a reverb-based preset isn't working for you.
• The three-band EQ doesn't change what you hear. Treat it as a visual staging area for now rather than something that reliably shapes the sound; reach for a voice preset (Bass Boost, Treble Boost) if you need an audible tonal change.
• Recording only shows up in the media library, not on a track. This happens when the video editor wasn't the active workspace at the moment you stopped recording. Open the media browser (Alt + 8) and drag the recording onto a track.
• A new subtitle track opens a file dialog you didn't expect. Adding a subtitle track from the track list always prompts for an .srt or .vtt file. Cancel the dialog if you meant to leave it empty, or run auto-subtitles instead, which creates the track for you.

Keyboard shortcuts

Shortcut	Action
Space	Play / pause, to preview your mix or captions
Ctrl + Z / Ctrl + Y	Undo / redo a mix or preset change

Mixer and preset controls are mouse-driven; see Timeline and tracks for the full transport and editing shortcut list.

Tips

Denoise before you mix

Clean up noisy dialogue with AI denoise before you start balancing levels; it is easier to judge a fader against a clean signal.

Let auto-detect run first

Try language auto-detect on the first pass of auto-subtitles rather than guessing the language yourself; it is right often enough to save the round trip.

Related

• Video editor overview for the full workspace
• Timeline and tracks for arranging audio and subtitle tracks
• Effects, transitions, and color for the visual side of a clip
• Video export for how audio and captions are included in the render
• AI features for the other on-device and AI-assisted tools in the editor