Diagnose the recording you already have
Keep the original OBS file unchanged and inspect a copy in an editor that can show its audio streams. Audition each track alone. Your ordinary player may only play one selected track, so hearing everything during playback does not prove that no other audio streams were saved.
Make a quick inventory: where is your microphone, where is the game, and where is the unwanted music? Record the content, not just the track number. A microphone-only stream may solve commentary recovery even if the game stream still contains a score. Conversely, two recorded streams may both contain the same desktop mix. Track count alone cannot establish that Spotify is isolated.
Stereo channels are also different from independent tracks. A two-channel stream usually describes left and right audio, not “microphone on one side and Spotify on the other.” Listen before discarding a channel. Removing one side of a stereo mix can change voices and effects while leaving the music audible.
For a general source-first workflow, see gameplay dialogue and effects. This guide focuses on the recording accident: choosing among already recorded streams, then deciding whether a mixed source is worth rescuing.
Why muting Desktop Audio can remove the game too
If game audio and Spotify were captured together, both are in the supplied mix. Muting that track can remove music, but also removes game dialogue, footsteps, alerts and other events. A clean microphone stream does not automatically include game sounds. Decide which components the finished edit needs before choosing a shortcut.
Future routing settings describe how new sound will be captured. They do not attach source labels to samples that were already combined. Similarly, changing the current Spotify volume or disabling Desktop Audio now affects the next recording; it cannot selectively alter the earlier file.
Case A: usable separate tracks already exist
Bring the recorded file into a suitable editor, identify the independent microphone and game streams, and leave out the unwanted music stream. Listen with the full mix disabled so that you do not accidentally include the song again through a second path. If one stream includes everything while others isolate sources, the combined stream is a reference, not an extra layer to stack on top.
Check whether in-game music remains in the retained game stream. An independent Spotify stream only isolates that application; it does not necessarily separate the game's own score from its effects and dialogue. If the game score is the problem, consider an authorized fresh capture with the game's music settings adjusted, or test separation on the mixed game source.
Review the reconstructed mix under picture. Check commentary timing, game events and the ending. A correctly chosen set of streams can still be assembled with an offset or doubled microphone. Preserve the original and export a clearly identified review version rather than overwriting the only capture.
Case B: microphone, game and music are mixed together
Choose a passage that contains all the things you need
Pick commentary under loud music and a necessary game event. Include a quiet word, a transient and its decay where possible. A menu screen with no speech may sound clean after processing while telling you nothing about the important scene. If commentary and action occur in different regions, record that one sample has not tested both.
The free preview is up to 15 seconds with the actual selection range shown by the tool. Short sources have less preview available. A long OBS session may also exceed the product's current file-size or duration limits; confirm those limits before preparing a working clip. This is not a batch-processing workflow for an entire recording library.
Choose a preservation goal based on the edit
Choose Dialogue & sound effects if your target includes commentary, in-game voices and scene sounds. Dialogue only focuses on people speaking and can intentionally reduce sounds you wanted. Neither goal guarantees every effect or word. The keep sound effects guide explains the listening tradeoff.
Use an authorized supported file in the music-removal tool. With video, the picture stays on your device while extracted audio goes to processing services. That distinction matters if a capture includes private conversation: a local picture workflow does not mean audio remains offline.
Compare commentary and game events independently
First follow your microphone speech. Can you still understand names, commands and short reactions? Then follow in-game dialogue if it matters. Finally inspect the specific game event you wrote down: a footstep pattern, an impact, a notification or the tail of an environmental sound.
Match timestamps and similar perceived voice loudness when switching between original and processed versions. Do not declare success from a waveform that looks smaller. Less audible score is useful only if the remaining content meets the edit's requirements. Listen to the pauses for pumping or music that returns between words.
Stop when the important sounds no longer work
A gunshot-like transient may resemble percussion. A tonal alert can resemble an instrument. A quiet line may sit under a singer in the unwanted music. Those overlaps can cause wanted content to disappear or unwanted content to remain. This guide does not assert a success rate for any of those sound classes.
If a required event changes meaning or a word disappears, return to the untouched source. Another pass on the separated result is not a dependable recovery technique. A partial edit, alternate capture, replacement narration or source assets may be a better route than forcing the soundtrack quieter.
When less music removal is the safer editing choice
If some original music is acceptable for your intended use, an editor can compare an original/processed blend. That can preserve a more natural scene, but also brings back music. It is not a solution when recognizable old music cannot remain. Current new previews offer preservation goals; do not assume there is a separate product “Reduce” control.
Approve the separated base before adding another score. New music may mask remnants but cannot reconstruct missing words or effects. If you plan to replace the bed, the replacement workflow covers that later stage. Review the actual quote before full processing, and inspect the complete exported result afterward.
Prevent the next recording accident
OBS's official multitrack guide describes assigning sources to recording tracks and enabling the required outputs. Separate capture sources and separate recording streams are both relevant: verify what each recorded stream actually contains rather than relying only on meter movement while recording.
For Windows setups, the official application-audio guide documents per-application capture and notes platform/version boundaries. It also warns about duplicate desktop capture when application audio is used. These features and their labels can vary by installed version; this guide has not yet reproduced the current UI on Windows or another platform.
Before an important session, make a short test containing microphone speech, game sound and a harmless test tone instead of copyrighted music. Open the resulting file in your editor and audition each stream independently. Write down the OBS version, operating system, source routing and recorded stream content. Keep the test private if it captures personal notifications or conversations.
A VOD-specific mix, a streaming mix and the local recording's audio streams should be checked separately. The fact that one destination sounds correct does not prove the local file contains the isolation you need later. This prevention advice does not recover an old recording.
Troubleshooting the saved file
Swipe sideways or use arrow keys while this table is focused to compare all columns.
| Symptom | First check | Avoid assuming |
|---|---|---|
| Only one track appears | Inspect streams in a capable editor or media tool | The player has displayed every stream |
| Commentary sounds doubled | Check whether full mix and isolated mic are both active | AI introduced every echo |
| Game sound disappears when music is muted | Audition the original desktop/game stream | Spotify was independently recorded |
| Speech is clipped before processing | Compare the untouched source | Separation can recreate missing peaks |
| Video is black but audio plays | Check original playback and codec support | Music removal caused a source-decoding problem |
| Sync drifts after export | Compare start, middle and end under picture | One aligned moment validates the full file |
Separate a technical failure from an unsatisfactory completed separation. If playback never starts, diagnose the file or player before judging audio quality. If it plays but a necessary event is absent, treat that as a content acceptance failure. The remedies are different.
For a finished video without source tracks, obtain the strongest original before trying another export. Repeated compression can make the scene harder to evaluate and cannot create independent production tracks.
Frequently asked questions
Can I recover every game effect exactly?
Do not promise that from a mixed soundtrack. Use source assets or a new authorized capture when exact event preservation is required. A preview tests only its selected interval.
How do I separate Spotify from Desktop Audio?
For an existing file, first check whether an independent application stream was actually recorded. If not, changing capture settings cannot retroactively split it. For new recordings, verify source routing using the official guide for your OBS version and platform.
What if the MP4 has only one audio stream?
One stream may still have several channels, but those channels are not necessarily isolated sources. Audition them and evaluate a separation preview only if you have a genuinely mixed source.
Does removing music clear copyright claims?
A local export does not automatically alter an existing platform upload or grant reuse rights. The YouTube decision guide explains the difference between a Studio claim edit and processing an authorized local source.
Can I process the entire OBS recording for free?
The free preview is a limited sample. File limits and the actual full-processing quote apply. See Pricing and the tool's validation; this guide does not promise unlimited processing.
Sources and testing limits
Documentation reviewed November 17, 2026. The OBS documents were checked that day and carry their own publication dates; documentation review does not establish a hands-on test of the latest installed software. This documentation-based guide has no installed-version screenshots or completed OBS routing test. It does not demonstrate a successful mixed-audio separation or full export. No personal capture, game footage or Spotify recording has been published as a case here. See the editorial policy.
