Mureka V9.5Mureka · Audio · per run
Songs with a female or male lead vocal, or instrumentals, in MP3, WAV or FLAC
- Your own lyrics
$0.27
Try3 models · from 84 credits
Mureka V9.5Mureka · Audio · per run
Songs with a female or male lead vocal, or instrumentals, in MP3, WAV or FLAC
$0.27
TryMureka Stem SeparationMureka · Audio · per run
Split a song into five WAV stems: vocals, instrumental, drums, bass and other
$0.11
TryMureka is the music model line from Skywork AI, and PopNow carries three of its models. Mureka V9 and Mureka V9.5 write music: you describe the style, add lyrics if you want singing, and get back a finished track. Mureka Stem Separation works the other way round. It starts from a finished song and splits it into separate audio files, so you can remix it, build a karaoke track or practise along with one part removed. Pick by what you have in hand, either an idea for a new song or a recording that already exists.
Both models take the same inputs. A style prompt of up to 1,024 characters describes the genre, instruments, mood, tempo and the kind of voice. Lyrics of up to 5,000 characters go in a separate field, and section tags such as [Verse] and [Chorus] tell the model how the song is laid out. Switch on instrumental and leave the lyrics empty to get background music with no singing. Results come back as MP3, WAV or FLAC, so you can choose a lossless file when the track is headed for a video edit or a mix.
The two models differ in price and in one setting. Mureka V9 costs less per song, which makes it a good place to try a brief and rework the wording. Mureka V9.5 costs more per song and adds an optional female or male vocal choice, for when the singer is part of the brief. On each model, a vocal song and an instrumental are charged at the same rate.
Mureka Stem Separation takes one song and returns five WAV files: vocals, instrumental, drums, bass and other. The instrumental is the full backing track, and drums, bass and other are that same backing divided three ways. It accepts MP3 or M4A files up to 10 minutes long, and the price is one flat amount per song, whether the song runs one minute or ten. Separation is an estimate rather than the original multitrack session, so reverb tails, backing vocals and instruments that share a frequency range with the voice can land in the wrong stem.
The models fit into one workflow. Write a song with V9 or V9.5 and keep the MP3, which is the default format and one that Stem Separation accepts. Then split it to get the vocal on its own for a lyric video, or the instrumental as a bed under a voiceover. Run only music you made or hold the rights to.
Start with Mureka V9 while you work on the style prompt and lyrics, since it costs less per song. Switch to Mureka V9.5 when you want to choose a female or male vocal. Both accept the same prompt and lyric limits, so one brief works on either model.
Yes. Turn on instrumental and leave the lyrics empty, and V9 or V9.5 builds background music from your style prompt alone. The price per song is the same as for a vocal track on that model. Describe the instruments, mood and tempo in the prompt, since there are no lyrics to guide the result.
Mureka lists up to about 5 minutes 30 seconds for a song and about 4 minutes 30 seconds for background music. The actual length depends on the brief and the lyrics: in our tests a folk song on V9 ran about 2 minutes 31 seconds, and a synth-pop song on V9.5 ran about 3 minutes 14 seconds.
Five WAV files, each as long as the song: vocals, instrumental, drums, bass and other. Add drums, bass and other together and you get the instrumental back. Keep the vocal stem for a remix, or use the instrumental as a karaoke track or a practice backing.
MP3 or M4A files up to 10 minutes long. WAV and FLAC files are refused before anything is charged, so convert a lossless master to a high-bitrate MP3 first. The price is one flat amount per song, so a ten-minute track costs the same as a one-minute track.
Only music you own or are licensed to remix. The result is also an estimate, not the original multitrack session, so listen to each stem before you build on it: backing vocals, reverb and instruments that overlap the voice sometimes end up in the wrong file.