MusicDesk

Every tool follows the same three steps. First, add your recording. The AI tools accept MP3, WAV, FLAC, OGG and M4A; the stem splitter, vocal remover and noise remover list a limit of 10 MB and 120 minutes per file, AI mastering accepts files up to 50 MB, and Audio to Text takes up to 10 MB and 20 minutes. Each workspace shows its own limits before you upload, so it is worth checking the tool rather than assuming they match.

Second, choose the result. Each workspace shows the options for its task: two, four or six stems or a dedicated instrument in the stem splitter, Recording noise or Audience sounds in the noise remover, Speech or Speech with music in Audio to Text, and an output format where it applies. For AI processing you sign in and review the estimated credits before submitting. Standard processing costs 0.1 credit per second of audio with a 6-credit minimum per task, billed by the second, and mode multipliers can apply. If your balance is too low, no task is created and nothing is deducted, and technical processing failures return the credits automatically.

Third, listen and download. Results open with a preview, often with a second track such as the isolated noise or crowd, so you can hear what was removed before keeping anything. Downloading an existing result again does not use credits, while running a result through another AI tool, for example separation followed by mastering, counts as a second billed task.

• •Vocal Remover and AI Stem Splitter - The vocal remover separates a song into a vocal track and an instrumental for karaoke, cover rehearsal or acapella edits. The stem splitter goes further with 2-stem, 4-stem (vocals, drums, bass, other) and 6-stem (vocals, drums, bass, guitar, piano, other) modes, plus dedicated single-instrument modes. The site is candid that these are estimated separations rather than the original studio tracks, and that overlapping sounds can appear in more than one stem.
• •Reference-Guided AI Mastering - AI Mastering takes your finished mix and a reference track whose sound you want to approach. The reference guides overall level, broad frequency balance, peak behaviour and stereo width, and the result page lets you switch between the original, the mastered version and the reference before downloading.
• •Voice Cleanup - The Background Noise Remover has a Recording noise mode for hiss, room tone and microphone noise and an Audience sounds mode that separates applause, cheering and laughter from live music, returning both the cleaned audio and the isolated noise or crowd track. A separate Reverb Remover reduces lingering room reverb in speech and vocals.
• •Transcription and Lyrics Extraction - Audio to Text detects the spoken language automatically and produces a transcript with timestamped segments you can replay, then copy or download as TXT; a Speech with music mode extracts the vocals first. The AI Lyrics Extractor recognises sung lyrics from a song, lets you edit text and line or word timing, and exports TXT, LRC, enhanced LRC, SRT, VTT, ASS and JSON files without further credits once recognition is done.
• •Audio to MIDI and MIDI Editor - A cloud Audio to MIDI task offers piano, melody and drum modes, while the free local converter runs in the browser. The free MIDI Editor creates or imports MIDI, edits notes, duration and velocity in a piano roll, adjusts tempo and instruments, and exports the result.
• •Free Browser Editors - The Audio Trimmer, Audio Joiner, Audio Converter, Audio Speed Changer and Slowed Reverb Generator need no credits and process audio in your browser. The converter exports to MP3, WAV, FLAC and other formats, and the site states that all of its export formats are free.

Musicians are the clearest fit. Singers can make backing tracks for practice or covers, instrumentalists can solo a guitar, piano, bass or drum part to learn it, and producers can pull stems for remixes or sample preparation, convert a melody to MIDI and test a reference-guided master on a finished mix. Because every AI task shows its credit estimate first and the free tools cover trimming and conversion, it suits occasional jobs as well as regular sessions.

Podcasters, interviewers and educators can clean background hiss or crowd noise from recordings, reduce room reverb and turn spoken audio into a timestamped draft transcript. Video creators can trim a music bed to length, join clips, speed audio up or slow it down, and create lyric or subtitle files from a song with the lyrics extractor.

The pricing shape makes it easy to start small: a free monthly allowance worth a few minutes of AI processing with MP3 output, one-time packs for occasional projects, and a monthly Pro plan for heavier users who need all six AI output formats, including 32-bit WAV and 24-bit FLAC. Anyone working on commercial releases should note that the terms leave rights clearance entirely to the user and that the audio-processing provider behind the AI tools is not named.