Write a song with AI, then split and mix audio online in one studio

It splits any song into vocals, drums, bass, guitar, piano and more, then keeps every part inside one browser studio. You can play synchronized stems, edit one part with A/B mutes and fades, set an A/B practice loop, mix with EQ and FX, detect BPM, musical key and chords, set loudness targets, export MIDI, change key or speed, clean noisy audio, and master or download the result.

Yes, on Studio. Describe the song and pick a length up to 5 minutes, choose instrumental or a male, female or duet voice, or use the Advanced tab to set the style, write the lyrics, choose the language, tempo, key and time signature. Or press Random: one button picks the style, the tempo and a 2 to 4 minute length and writes the lyrics, and you choose only the voice, the language and whether to split it. Two versions come back for one charge; keep the one you like and it becomes a normal track you can split into stems, mix, master and export. Songs are charged by length from HD and AI minutes and refunded if generation fails.

Yes, on Studio, once the track is separated into stems. Write your own lyrics or, when that option is on, let the AI write them. Pick a voice, and the AI sings over your other stems while the original vocals are muted. Two takes come back for one charge; keep one or both as ordinary stems, or ask for more takes of the same lyrics (a new request, charged the same way). Each request is charged by the track's length from HD and AI minutes and refunded if generation fails. Lyrics you supply must be yours; lyrics the AI writes are generated output.

Yes, on Studio, once the track is separated into stems (an AI song you kept counts too). Choose Backing chorus on the AI lyrics panel, pick up to four sections of the song typed to the millisecond or taken from your A/B loop, choose a male, female or mixed choir, and choose what it sings: lyrics you write, the song's own hook when the track is an AI song, or wordless ooh and aah. The choir can hear your lead singer while it sings, for harmonies around the voice, or only the band. Your original vocals stay and the chorus arrives as one new stem, silent outside the sections you chose. Two takes come back for one charge, charged by the total length of the sections from HD and AI minutes and refunded if generation fails.

Yes, on Studio. Under the style box press Write it for me (AI): describe the song in a few words, or paste a YouTube link and the style line is written from the video's title (only the title is read, no audio is downloaded). Under the lyrics box press Write lyrics (AI): say what the song is about and tagged lyrics come back in the language you chose. Both go into the box in front of you to edit, both are written by an AI, and neither uses HD and AI minutes. Each account has a daily allowance.

Cancel any time from the billing portal: no emails, no phone calls. You keep your plan until the end of the period you already paid for. After that your account is paused and your files are kept for 30 days so you can resubscribe and pick up right where you left off. If you don’t resubscribe within 30 days, your uploads and stems are permanently deleted and your account becomes a free account you can still use. Deleting your account instead removes everything (uploads, stems and profile) immediately.

Yes, on Studio. Open My voice on your Account page, or choose Use my voice on the AI lyrics panel, and record 10 to 30 seconds of your own singing, or upload a file, tick three boxes and read a short phrase aloud into your microphone; the check confirms the reading matches your sample and records your consent. Then any AI lyrics request can be sung in your voice instead of an AI voice, and any vocal stem of a track, an AI take or a kept AI song, can be sung back in your voice as a new stem called My version, with the original muted. Only your own voice can be used, it is never used to train AI, and you can delete it at any time. It uses HD and AI minutes, charged on top of the take, and a request that delivers nothing is refunded.

Yes, on Studio, once the track is separated into stems. Then open the vocals stem's menu, choose Switch to a high or low tone and pick High tone (female), Low tone (male) or Auto, which measures the singing and picks the opposite voice. A stock AI voice of the other gender sings the stem again with the same words, melody and timing, moved by whole octaves so the key of the song does not change, and it arrives as a new stem while the original is muted, never deleted. The stock voices are AI-made, not recordings of a real person. It uses HD and AI minutes, charged once by the stem's length, and a request that delivers nothing is refunded.

Yes. Every separated stem stays synchronized in the browser mixer and has its own volume, Mute, Solo, 10 band EQ and effects rack. You can mute or fade a selected range of one stem without changing the others. Studio can export the custom mix with your stem edits, EQ, key and effects choices included.

Yes. Set A and B, then mute that range on one stem, add a fade in or fade out, restore the selected range, or clear every edit on that stem. The original stem is not rewritten. Playback, custom mixes, sharing, loudness normalization and HD mastering follow the edits you made.

Yes. Make A/B loops, mute or solo parts and seek on synchronized waveforms. Studio adds automatic BPM, musical key and chords with optional changes between beats, Auto Metronome tick speed, key control and playback speed that preserves pitch.

Yes, on Studio. Vocal tools include reverb, delay, doubler and a Harmonizer with Deep and Light voices. Studio hears every effect live and can include it in individual lossless WAV stems and custom mixes. Free and Starter can open every rack before upgrading.

Upload MP3, WAV, M4A, FLAC, OGG and other common audio formats. Tracks up to 90 minutes are supported. Every paid plan downloads MP3. Studio adds lossless WAV and custom mixes.

No. Your audio is processed solely to produce your stems, is never used to train or improve any model, and is never sold or shared with model vendors.