To export only the audio, open the Deliver page and double-click the Audio Only preset, or start from Custom Export and clear Export Video. In the Audio tab, set Format to WAVE for a master or MP3 for a small listening copy, and point Output Track at the bus that carries your whole mix. Click Add to Render Queue, then Render All. For video post I deliver 24-bit/48 kHz WAV unless the delivery specification asks for something else.
Dialogue, music and effects stems need one extra step: route each group to its own Fairlight bus first, then render every bus over the same timeline range.
Decide What You Are Delivering: One Mix or Stems
Before you open the Deliver page, answer one question: should the person on the other end hear your finished mix, or do they need to rebalance parts of it?
| Deliverable | Choose in Resolve | What it contains |
|---|---|---|
| Finished stereo mix | Main stereo bus | All tracks, automation, track processing, and bus processing routed into that bus |
| Dialogue, music, and effects stems | Separate D, M, and E buses | Processed groups that can be recombined downstream |
| Raw or lightly processed tracks | Individual timeline tracks | Each selected track, which may not include downstream bus processing |
| Bounced clip or track | Fairlight bounce command | A rendered asset created during editing, not necessarily the final delivery master |
| Separate clip files | Fairlight Export Audio Files, or Individual clips on Deliver | One file per clip in the chosen range |
| Pro Tools or DAW session | Pro Tools preset on Deliver | A Pro Tools AAF with linked or embedded audio, track names and volume automation, but no audio effects |
For approval files, podcasts, YouTube voiceovers and final audio masters, use the approved Main or Submix bus unless the handoff specification says otherwise. A timeline-track export isn't automatically a stem. If dialogue passes through a dialogue bus with EQ or compression, the bare source track can sound noticeably different from the mix you signed off.
Blackmagic’s Fairlight overview covers the routing model. The Resolve 21.1 Reference Manual defines the Deliver Audio panel: Output Track picks a Main or Submix bus, the plus button beside it adds more outputs to the same job, and All Timeline Tracks writes each timeline track as its own track in the file. That last choice gives you separate tracks, with no processed stereo mix among them.
How to Export One Audio File from DaVinci Resolve
- Set the render range. On the Deliver page, choose Entire Timeline unless you set In and Out points on purpose. Leave room for the last reverb or music tail.
- Choose Audio Only or Custom Export. Audio Only is a starting point, and its defaults catch people out: the 21.1 manual says it renders the Main 1 bus as a single clip, one track per channel, in MXF OP-Atom with Linear PCM at 16-bit. For a stereo WAV master you'll change most of that in the next steps. Custom Export shows every setting from scratch. Give the file a useful name such as
project_mix_v03.wavand save it to a normal delivery folder, never a cache directory. - Confirm that Export Video is disabled. Audio Only does this for you. In Custom Export, open the Video section and clear Export Video; Resolve then disables the video controls and shows an Audio Format menu. If the deliverable needs picture, follow the video-export workflow.
- Enable Export Audio. In the Audio section, switch on Export Audio.
- Choose WAV and Linear PCM. Set the format to WAVE and the codec to Linear PCM, then pick 24-bit and the project's sample rate. In video post, that's normally 48 kHz. Depending on the format, the Codec menu can also show IEEE Float, MP3 or AAC where the selected format and platform offer it. If someone needs a 16-bit file later, make it from this 24-bit master as the very last conversion and dither it once there.
- Select the real output bus. In Output Track, choose the stereo Main or Submix bus that carries the approved mix. Go by the name the project gives that bus; Resolve doesn't guarantee that every session calls it Bus 1.
- Keep channel-splitting options off for a normal stereo master. The Audio Only preset starts with Render one track per channel switched on, so clear it. Render as discrete audio tracks also splits channels into separate tracks and drops their position information. Turn either one on only when the delivery specification requires that structure.
- Add and render. Click Add to Render Queue, read the summary of the queued job, then click Render All.
If Bus 1 isn't really your mix bus, stop and fix the routing before you render anything. Open the Fairlight mixer, play the timeline and watch the bus meters. Whatever bus you pick for delivery has to receive every track you expect to hear. The Fairlight bus-routing guide traces the track-to-bus signal path, so you can see exactly where a stray track drops out.
WAV, MP3, or AAC?
WAVE with Linear PCM is the safe default. It stores uncompressed PCM audio, so another edit or encode doesn't stack lossy-compression damage on top, and the channel count, sample rate and bit depth are all explicit in the file. Unless a broadcaster, editor or client hands you a different specification, 24-bit, 48 kHz WAVE/Linear PCM is the common video-post starting point.
MP3 and AAC are listening-copy formats. They're small and handy for review. Don't let one stand in for an uncompressed master that somebody will edit, mix or encode again. The Format menu also lists FLAC, which is lossless and smaller than WAV; send it only when the recipient has said they accept it.
The 21.1 codec list includes AAC encoding for .aac and .m4a on Windows and macOS, with no AAC support listed for Linux. The Audio panel description in the 21.1 manual still says encoding is macOS-only, so those official references disagree. I would use the platform-specific codec table for compatibility and check the actual Format and Codec menus before promising an AAC delivery; neither document establishes that every preset exposes the same choices.
| Format | Use it for | Avoid it when |
|---|---|---|
| WAV / Linear PCM | Master, archive, editor handoff, stems | File size is the only concern |
| MP3 | Small speech or music review copy | The file will be processed or encoded again |
| AAC (Windows/macOS; check format) | Compressed listening copy when the chosen format offers AAC | The recipient asked for an uncompressed master |
How to Export Dialogue, Music, and Effects Stems
A stem is a deliberately routed group, and the word doesn't stretch to cover every visible timeline track. A Dialogue bus can carry the intended EQ, compression, automation and level trim, while a source-track export can skip all of that downstream bus processing.
Check the actual routes and insert states first. A bus export holds exactly what reaches that selected output path, and that can differ from what the editor meant to include.
- Create or identify the processed stem buses. Route dialogue to DX, music to MX and effects to FX, in whatever channel format the handoff calls for.
- Verify the routes and processing. Watch each bus meter, check inserts and automation, and make sure the buses reach the approved monitoring path without double-counting.
- Set one common start and end range. Every stem must start at the same moment and run past the same final tail, including stems that sit silent at the head or tail.
- Create one render job per standalone stem file. Pick only the intended bus in Output Track, give it a unique filename such as project_DX_v03.wav, and add the job to the queue. Repeat for MX and FX without touching the range, sample rate, bit depth or format.
- Render and verify. Reimport all the stems at the same start time, sum them at unity gain and compare the result with the approved mix. A limiter or other processing on the Main bus itself sits after the stem buses and isn't in the stem files, so expect the sum to differ wherever it acts.
On bigger sessions, set the groups up well before the last day of the mix. Buses bolted on at export time are exactly how sends, sidechains and bus plug-ins get missed. If dialogue dynamics are part of the approved sound, check the routing alongside the dialogue compressor workflow.
How to Export Each Audio Clip as a Separate File
Sometimes the person on the other end wants the pieces, not the mix: every interview clip as its own file, or each music cue cut where it sits in the edit. On the Fairlight page, select the clips in the Timeline, right-click one of them and choose Export Audio Files. The Fairlight export section of the manual walks through the dialog in order: Browse for a destination, optional name options, then File Format, Sample Rate, Bit Depth and a Channel Format of Multi-Mono or Interleaved. Click Export when the settings match the request.
Source offers Individual Clips or Consolidated Clips, and a dropdown sets the range: the complete timeline, a selected range or just the selected clips. Checkboxes limit the job to selected tracks, include hidden tracks and pad files to frame boundaries. The include options cover Clip Level, Clip Fades, Clip EQ and FX and Clip iXML, and the dialog can normalize to the same loudness standards the Fairlight page offers. Track and bus processing aren't on that list, so I'd keep this route for raw clip material and render anything that has to sound like the mix from a bus on Deliver.
Deliver can do the same job from the render queue. Start from the Audio Only preset and set Render to Individual clips. The manual's Single Clip vs. Individual Clips section says each clip in the range becomes its own media file, with timecode cloned from the original source media. In the File panel, Use Source Filename keeps each source file's name, and Use unique filenames, available with a custom name, adds the track and clip number so two clips cut from the same source don't overwrite each other.
Bounce Is Not the Same as Deliver
Fairlight bounce commands make working media during the edit. The Fairlight bounce section of the 21.1 manual says Bounce Mix to Track processes a selected bus mix onto a new or existing timeline track and writes the new media to the folder set in Project Settings > Capture and Playback > Save clips to.
Deliver is where the controlled handoff gets built: job, range, filename, destination and output assignment.
Bounce Mix to Track suits a reusable asset on the timeline; Deliver is for the final master or handoff. A bounce that worked tells you nothing about the Deliver range, the selected output, the channel layout, normalization or the destination.
How to Send Audio to Pro Tools or Another DAW
When a mixer will finish the sound in Pro Tools, pick the Pro Tools preset on the Deliver page. The Pro Tools preset entry in the manual says it writes a specially formatted AAF, the linked audio files and a linked reference video, for Pro Tools or any application that imports a Pro Tools style AAF. Send the AAF the preset writes next to the media. File > Export AAF/XML won't give Pro Tools the correct exchange file, and the Avid AAF preset is meant for a return trip to Media Composer, not for Pro Tools.
The Codec menu in the Audio panel sets the package. Linear PCM writes individual audio files linked to a separate AAF, and Embedded in AAF puts Broadcast WAV audio inside the AAF as a single deliverable. Render one track per channel is on by default, so a 5.1 polyphonic WAV leaves as six mono WAV files. Clear it only after checking that Pro Tools supports the multichannel format you plan to send. Every channel of the source media goes out, including channels muted in Clip Attributes.
A few more settings shape what the mixer receives. Add X frame handles, in the Advanced Settings of the Video panel, adds extra frames to the start and end of each clip so the mixer has room to adjust edits. The video format sets the audio file type: MXF OP-Atom gives MXF audio and QuickTime gives Broadcast Wave. Exported files keep the source media's name and timecode, and clips you synced in Resolve take the name and timecode of the synced audio file. Track names, all track and clip volume automation with its keyframes, and iXML metadata with channel names travel too. The reference movie renders with effects and titles; clear Export Video in the Video panel if you don't need it.
Audio effects don't make the trip. The manual's Pro Tools export warning says FairlightFX, EQ, Compression, Pitch and Elastic Wave effects are neither exported nor baked in. If the AAF misbehaves in the DAW, look for audio effects or audio compound clips in the timeline and replace them with duplicates of the same clips without effects. The compound clips guide shows how Decompose in Place breaks a compound clip back into its parts. I'd also render a stereo WAV of the Main bus and send it with the AAF, because the AAF carries none of your EQ or noise cleanup and the mixer needs to hear what the client approved.
Check Loudness, Peaks, and the Render Range
A quiet export is no reason to switch on normalization. Measure the mix first, find out what loudness target the delivery needs, and adjust the mix on purpose. A web video, a podcast, a broadcast master and a cinema deliverable can all ask for different things. The Resolve loudness and LUFS guide explains what to measure before you lean on any target-based normalization.
If a target does apply, the Deliver page has its own controls. The Audio Normalization settings in the manual sit in the Audio tab and work per output: Normalize to Standard scales loudness or peak, whichever crosses its threshold first, while Optimize to Standard keeps loudness in the desired range with peaks under the threshold. Pick the Standard, set Target Level in dBFS and tick the Output Track it applies to; unchecking Audio Normalization bypasses it. I'd leave it off on a master going to a mixer or broadcaster, because normalizing on export hides whether the mix itself hit the target. On a listening copy for a platform with a known target, it's a reasonable shortcut.
- Start: make sure the first transient or word doesn't sit before the chosen In point.
- End: listen past the last visible edit so reverb, room tone and music decays don't get chopped.
- Peaks: fix clipping in the mix itself. Saving the same clipped signal as WAV won't repair it.
- Channels: check mono, stereo or multichannel delivery against the specification.
- Version: put the revision in the filename. Overwriting an approved file that carries a vague name such as
final.wavis how the wrong version reaches the client.
Resolve Free vs Studio
The standard audio-only Deliver workflow, WAVE/Linear PCM export, Fairlight mixing and ordinary bus routing all work in DaVinci Resolve Free. Specialized codecs and advanced workflows can vary with the operating system, container, installed components and edition.
Edition labels won't tell you what a given machine can render, so open the actual Audio panel and look. Codec availability depends on the operating system and selected format. The platform-specific 21.1 codec table lists AAC encoding on Windows and macOS, with no AAC support listed for Linux; check the actual Format and Codec choices before committing to delivery. When the recipient hasn't sent any other specification, WAVE/Linear PCM is the portable, dependable choice for an audio-only master.
Fix Common Audio Export Problems
The exported file is silent
Confirm that Export Audio is enabled and that the selected bus meter moves during playback. Then go through track mute, bus mute, automation and routing. If the timeline itself plays back silent, start with the Resolve no-audio checklist before you change render settings.
The export contains only one track or misses music
Most likely you picked a timeline track, or a bus that doesn't receive the whole mix. Follow the signal from each source track through any subgroup bus to the output chosen on Deliver. In a project with custom routing, Bus 1 isn't necessarily the master, and the Audio Only preset's default of Main 1 may not be the bus you mixed into.
The WAV is mono instead of stereo
Look at both the bus format and the Output Track selection. A mono bus gives you a mono file no matter how many audio clips sit on the timeline. Change the routing or create the stereo bus you need; renaming the file won't change its channel layout.
The render has separate tracks or channels instead of one stereo mix
Go back to Audio and select the intended stereo Main or Submix bus. Disable Render one track per channel and Render as discrete audio tracks unless the delivery specification requests that structure, and remember that Audio Only switches the first one on by default. Those switches change how channels are stored in compatible rendered media. They don't create a processed stereo bus or guarantee one standalone WAV file per stem.
The file ends too early
Check the In/Out range and move the Out point past the full audio tail. Entire Timeline is the safer choice for a normal master, but listen through the last few seconds anyway, because a muted or trimmed tail can still end abruptly.
The render is quieter or different from playback
Make sure Deliver uses the same bus path you monitored and that its meter moves across the full range. A bus render includes only the audio routed through that output, with its active insert and automation state. A timeline-track output can sound different, and normalization in the render job can change the level after the mix.
A Two-Minute Verification After Every Export
Open the exported file in a player outside Resolve and listen to the beginning, a loud section, a quiet section and the ending. Check the duration and the channel count. For a client master or stems, reimport the files into a clean timeline, line them up at the same start time and compare them with the approved mix.
A green render-complete message only proves that Resolve wrote a file. It says nothing about whether you chose the correct bus, covered the full range, kept the channel layout or delivered the version the recipient asked for.
Once that check passes, keep the WAV master and make a smaller listening copy only when somebody asks for one. For the rest of the picture-and-sound handoff, the DaVinci Resolve export settings guide picks up from here.
The routing, dialogue, loudness and bus work that comes before an audio master lives in the Fairlight audio hub. This page sticks to audio-only files, buses, tracks, stems and verification.
DaVinci Resolve Audio Export FAQ
What are the best settings for audio-only export?
For a general video-post master, use WAV, Linear PCM, 24-bit, 48 kHz and the main stereo bus. If a supplied delivery specification asks for something different, follow it.
How do I export one finished stereo mix?
Turn off Export Video, turn on Export Audio, and choose the stereo bus that receives the complete approved mix. Keep the individual-track and split-channel options disabled; the Audio Only preset switches Render one track per channel on by default.
Can DaVinci Resolve export separate audio stems?
Yes. Route dialogue, music and effects to separate processed buses, then create one named render job per bus with an identical start and end range, format, sample rate and bit depth. Reimport and sum the files once to verify alignment and content.
Why is my exported audio silent?
The usual causes are Export Audio being off, the wrong output bus, a muted bus or track, automation, or source tracks that never reach the selected bus.
Should I use 44.1 or 48 kHz?
Match the project and the delivery specification. Video post-production normally uses 48 kHz; some music-only workflows use 44.1 kHz.
Does audio-only export work in Resolve Free?
Yes. Resolve Free handles the standard audio-only workflow, WAV export, mixing and ordinary bus routing. Specialized formats and advanced workflows can depend on Studio, the operating system or additional components.
