How to extract a single channel from stereo audio
- Step 1Check the file is stereo — Channel extraction only makes sense on a 2-channel file. After adding the file, the tool shows
2chormonoin the file row. If it readsmono, there is no separate left/right to pull apart. - Step 2Add your stereo file — Drop in
.wav,.mp3,.m4a,.flac,.ogg, or.opus. FFmpeg 8.1 runs in WebAssembly inside the page — nothing uploads. - Step 3Run the split — The tool runs
channelsplit=channel_layout=stereoand produces two mono files. There are no parameters to set — it always returns both channels. - Step 4Identify which file is the channel you want — File
-1is the left channel; file-2is the right channel. If you are unsure which side holds the part you want, listen to both — they are short to scrub through. - Step 5Download only the channel you need — Use the per-file Download link for the single channel you want, and ignore the other. Or use Download all and delete the unwanted file.
- Step 6Clean up the extracted channel — Run the kept mono WAV through ai-noise-reducer for room noise, voice-eq for tone, or loudness-normalizer to set a delivery level.
When channel extraction works
Channel extraction copies one side verbatim. It only isolates a part if that part lives on its own channel.
| Goal | Works with this tool? | Why / what to use instead |
|---|---|---|
| Keep one mic from a dual-mono recording | Yes | The good mic is the whole content of one channel — extract it cleanly |
| Isolate a vocal hard-panned to one side | Yes | If the vocal occupies a single channel, that channel is the vocal |
| Remove a centred lead vocal (karaoke) | No | A centred vocal is equal in both channels; needs phase cancellation, not extraction |
| Get a mono mixdown of a stereo song | No | That is a downmix (L+R), not one channel — extraction keeps only one side |
| Check the right channel for a dropout | Yes | Extract the right channel (-2) and listen to it alone |
Extraction behaviour and output
Real facts from the processing code — fixed behaviour, no options panel.
| Property | Value |
|---|---|
| Filter | channelsplit=channel_layout=stereo (FFmpeg 8.1 WASM) |
| Required input | Stereo (exactly 2 channels) |
| Outputs | Two mono files — left = -1, right = -2 |
| Format | 16-bit PCM WAV (pcm_s16le), original sample rate |
| Selective extract? | No — both channels are produced; you keep the one you want |
| Upload | None — runs in-browser |
Tier limits
Audio family. Size and duration caps are enforced independently.
| Tier | Max size | Max duration | Files/run |
|---|---|---|---|
| Free | 50 MB | 30 min | 1 |
| Pro | 200 MB | 120 min | 10 |
| Pro + Media | 100 GB | Unlimited | 100 |
| Developer | 100 GB | Unlimited | Unlimited |
Cookbook
Extraction recipes that work — and the matching 'use a different tool' cases. Left channel is always file -1, right is -2.
Keep only the left channel (good lapel mic)
A dual-mono recording where the lapel mic on the left is clean and the right channel is a dead handheld. Extract the left, drop the right.
Input: presser.wav (2ch, lav=L good, handheld=R dead) After split: presser-channel.wav-1 <- keep (left, lav) presser-channel.wav-2 <- discard (right, dead) Keep the -1 file only.
Extract the right channel for a balance check
You suspect the right channel has an intermittent dropout. Pull it out and audition it on its own.
Input: master_v3.wav (2ch) Keep: master_v3-channel.wav-2 (right channel, isolated) Listen for the dropout in isolation, then fix at the source.
Hard-panned guitar isolated for re-amp reference
A rough mix has the rhythm guitar panned hard right. Extracting the right channel gives a usable mono reference of just that guitar.
Input: roughmix.mp3 (2ch, gtr panned hard R) Keep: roughmix-channel.wav-2 (right = guitar-dominant) Note: anything else also panned right will be present too.
What NOT to use it for: centred vocal removal
A pop song with the lead vocal in the centre. People expect 'extract a channel' to make a karaoke track — it can't, because a centred vocal is identical in both channels.
Input: song.mp3 (2ch, lead vocal centred) Extract left -> still has the vocal Extract right -> still has the vocal Why: centred = equal L and R. Channel extraction keeps the whole side. True karaoke needs phase-cancellation, not this tool.
Salvage one side of a phone-call recording
A two-track call where your mic (left) is clean and the far end (right) is codec-mangled. Keep your side as a clean mono WAV for a quote.
Input: call.m4a (2ch, you=L, them=R) Keep: call-channel.wav-1 (your side, clean WAV) Then run voice-eq or ai-noise-reducer on it.
Edge cases and what actually happens
Trying to remove a centred vocal
Not supported hereChannel extraction cannot create a karaoke/instrumental track from a normal mix. A centred lead vocal sits equally in both channels, so neither extracted channel removes it. That requires a phase-cancellation (vocal-removal) process, which this tool does not perform.
Input is mono
Rejected (requires stereo)There is no left/right to extract from a single channel. The tool throws Channel splitter requires a stereo input. for any mono file.
Wanting a mono downmix instead of one channel
By designExtraction keeps one side only; it does not sum L+R into a mono mixdown. If a stereo song has content spread across both channels, taking only one channel loses whatever was panned to the other. A true mono downmix is a different operation.
Both channels sound the same
By designIf the source is dual-mono (identical L and R) or a centred-mono-on-stereo file, both extracted files will be identical. That is faithful — the channels genuinely held the same audio. Keep either one.
Multichannel input (5.1)
Rejected (requires stereo)Only exactly-2-channel input is accepted. A surround file is rejected with the stereo-required error; this tool does not pull a single channel out of a 5.1 layout.
Output is WAV, not the source format
ExpectedThe kept channel is always written as 16-bit PCM WAV regardless of input. There is no format option. Convert with wav-to-mp3 if you need MP3.
Hard-panned part has spill from other instruments
By designExtracting the right channel gives you everything on the right, not just the one part you wanted. If other elements were also panned right, they come along. Extraction is channel-faithful, not source-separation.
Long file over the Free duration cap
413 tier limitFree allows 30 minutes per file; longer files need Pro (120 min) or higher. The duration cap is separate from the 50 MB size cap, so a long compressed file can pass on size but fail on duration.
Frequently asked questions
Can I extract just the left channel without also generating the right?
The tool always produces both channels as separate mono files — left is -1, right is -2. To 'extract only the left', simply download -1 and ignore -2. There is no selective single-channel button.
Can this make a karaoke / instrumental track?
Only if the vocal is hard-panned to one channel — then extract the other channel. For the usual case of a centred lead vocal, no: a centred vocal is equal in both channels, so neither extracted side removes it. That needs phase-cancellation vocal removal, which this tool does not do.
Which file is left and which is right?
File -1 is the left channel and file -2 is the right channel. If you aren't sure which side holds the audio you want, listen to both before deleting one.
What format is the extracted channel?
Mono 16-bit PCM WAV at the original sample rate. There is no format chooser. Run wav-to-mp3 afterwards if you need a compressed deliverable.
Is the extracted channel exactly the same as that side of the original?
Yes — channelsplit copies each channel's samples verbatim, so the kept WAV is a sample-exact mono copy of that channel (re-containerised as PCM). No level change, pan, or downmix is applied.
Why does the channel I extracted still contain instruments I didn't want?
Because they were panned to that channel too. Extraction keeps the whole side, not an isolated instrument. It is not source separation — it can only give you what was already on that one channel.
Can I get a mono mixdown (L+R combined) instead?
Not with this tool — it keeps one channel, it does not sum the two. A mono downmix is a different operation; extracting a single channel discards whatever lived on the other side.
Does it work on a mono file?
No. A mono file has one channel, so there is nothing to extract. You'll get Channel splitter requires a stereo input. For width on a mono file, use mono-to-stereo.
Is anything uploaded?
No. FFmpeg 8.1 runs in your browser via WebAssembly; the audio stays on your device — suitable for confidential or unreleased material.
What input formats are accepted?
WAV, MP3, M4A/AAC, FLAC, OGG, and Opus — provided the file is stereo. FFmpeg decodes the container, then splits.
How long a file can I process?
Free allows 30 minutes and 50 MB per file; Pro raises this to 120 minutes / 200 MB, and Pro + Media removes the duration cap (100 GB).
Can I extract a channel, clean it, and use it as a standalone clip?
Yes — that's a common flow. Extract the good channel, then run ai-noise-reducer and voice-eq on the mono WAV, and you have a polished single-source clip.
Privacy first
Every JAD Audio tool runs entirely in your browser via FFmpeg (WebAssembly) and RNNoise. Your audio files never leave your device — verified by zero outbound network requests during processing.