2026.10.14
INMU KING audio edits: replace words in the original song
Use the original INMU KING track and replace vocals in any target passage: repeated names, verse lyrics, chorus lines or shouts. Change one word, a line or a whole verse. Keep the backing and select the region you want to edit.
This page contains affiliate links.
Examples include replacing name calls with your name, rewriting verse lines about a person or event, and changing shouts. “お前寝不足スギ” is just one short example. For any passage, match its onset, ending and held notes. Record long verses line by line and assemble them; record words absent from the source yourself.
- This workflow uses Windows and Audacity 3.7.1. The official OpenVINO Windows link points to a 3.7.1 installer. Use matching application and plugin versions, not the latest Audacity.
- Run the installer with the Music Separation models included, wait for downloads and restart Audacity. Confirm Effect → OpenVINO AI Effects → OpenVINO Music Separation exists before proceeding. If missing, correct the version pairing first.
- Put the permitted source MP3 or WAV in your work folder as source.mp3 or source.wav. If you have only a permitted local video, import it into CapCut, put it on the timeline, right-click → Extract audio, then choose audio export in Export. Start with a local file.
1. Separate backing and vocals
- Open Audacity and use File → Import → Audio to select the source. Save the project as inmu-edit.aup3 using File → Save Project → Save Project. The aup3 is an editable project, not the audio for posting.
- Select the full song with Ctrl+A, run Effect → OpenVINO AI Effects → OpenVINO Music Separation in 2-Stem mode. Instrumental and Vocals appear. Separating the full song keeps both aligned with the source start.
- Mute the original source track. Leave Instrumental and Vocals unmuted and turn every Solo off. Press Space to hear their combined result. Playing the source too would double the audio.
- Solo Vocals and find the name call, lyric line or shout to replace. Stop, drag-select its waveform and zoom with Ctrl+mouse wheel. Replay with Space and include the complete onset and ending. Retain the old vocal for alignment.
Audacity OpenVINO · Music Separation
2. Record replacement words
- Open a separate Audacity window with Ctrl+N. In Audio Setup, set Recording Device to your microphone and Recording Channels to 1 (Mono). Use headphones, not Stereo Mix or system playback as the input.
- Listen to your selected passage, then press red Record in the recording window. Sing the replacement name, lyric or shout on the original beat and pitch; Stop and replay with Space. Record a whole verse line by line and save each line separately. Lower input gain if it clips.
- Select and Delete long surrounding silence, preserving first consonants and final tails. Use File → Export Audio → Export to computer to save WAV as replacement.wav. For several lines, use replacement-01.wav, replacement-02.wav, etc. Import each through File → Import → Audio in the source project.
- Optionally open ElevenLabs Voice Changer, drag replacement.wav into its audio upload area, choose the target voice and Generate. Download your selected result as replacement-ai.wav or replacement-ai.mp3. Import either the recording or conversion into Audacity, not both.
Voice Changer is available on Free: about 1,000 credits/minute, or ten minutes if all 10,000 monthly credits are used here. This is separate from TTS character billing. Credits are shared across features; free output is non-commercial and requires attribution. Commercial use starts with Starter, normally $6 billed monthly.
ElevenLabs Voice Changer · Checked 2026-10-27
ElevenLabs Voice Changer ↗ · Voice Changer guide
3. Replace on the original beat
- Show both original and replacement waveforms. Drag the new clip’s top name bar left or right to align its first syllable onset with Vocals. Dragging inside the waveform selects audio; use the header bar to move it.
- Select only the target region on the Vocals waveform and press Ctrl+L. It becomes flat: silence without removing time. Do not Delete, which could shift later vocals. Do not select the backing track.
- Keep the source muted and turn Mute/Solo off for the working tracks. Play from before the edit and align its first sound, internal beats and ending using the clip header. If only the start matches, re-record at the correct length. Align multiple lines separately, then play the whole verse.
- Match replacement level to neighboring vocals using its track Gain slider. For a click at a join, select only a very short opening edge and apply Effect → Fading → Fade In; use Fade Out at the end. Do not fade the entire word.
- Play the mix and check for doubled lyrics. Ctrl+S saves the project. In File → Export Audio → Export to computer, choose WAV, Stereo, 44,100 Hz and Signed 16-bit PCM, with Export Range set to Entire Project. Save inmu-collage.wav, not just the selected region.
For a single-word edit, retain surrounding vocals and silence only that small region. For a line or verse, silence the matching region and place each recorded line at its original position. Align repeated name calls individually; one aligned repetition does not finish the whole passage.
4. Add the edit to the dance clip
- Create a CapCut project and Import the dance video and inmu-collage.wav. Drag video onto the main track and WAV onto an audio track. Select the video and set Audio → Volume to 0; keep the WAV audible.
- Trim the audio to the chorus and move its clip to match the first dance movement. Play through and Export as MP4 at 1080p. Open the saved file to check the replaced words, audible audio and dance start.
Use source audio permitted for editing and publication. These documentation-based steps have not been tested with INMU KING audio separation or conversion.