TTS Input Text
0/10000
Voice Selection
TTS Generated Audios

Your generated audio will appear here

Enter some text on the left and click Generate to get started.
Voice Cloning Input
Voice Clone Generated Audios

Your cloned voice will appear here

Fill in the form on the left and click Clone & Generate.
Voice Design Input

Longer, detailed descriptions give better voices: age, gender, accent, pace, mood and use. Example: “a warm British woman in her 30s, slow and friendly, for audiobooks”.

Voice Design Generated Audios

Your designed voice will appear here

Describe a voice on the left and click Design & Generate.
Voice Changer Input
Voice Changer Generated Audios

Your converted audio will appear here

Fill in the form on the left and click Change Voice.
AI Dubbing Input
More options
AI Dubbing Generated Results

Your dubbed video will appear here

Choose a video on the left and click Generate Dub. The new voice replaces the original sound.
Speech to Text Input
Speech to Text Generated Results

Your transcript will appear here

Fill in the form on the left and click Transcribe.
Music Generator Input

Longer prompts give better music: genre, mood, instruments, tempo and where it will be used. Example: “calm lo-fi hip hop, soft piano, light rain, 80 bpm, for studying”.

More options
Music Generated Audios

Your music will appear here

Describe a track on the left and click Generate Music.
SFX Generator Input

Detailed prompts give better sounds: one clear sound, what makes it, the place and the distance. Example: “single wooden door knock, close up, small dry room”.

SFX Generated Audios

Your sound effect will appear here

Describe an effect on the left and click Generate SFX.
Song Generator Input

Describe the vocal, genre, instruments, tempo and mood for the best song. Example: “soft female vocal, acoustic pop, guitar and piano, 90 bpm, warm and hopeful”.

More options
Song Generated Audios

Your song will appear here

Fill in the form on the left and click Generate Song.
Lyrics Generator Input

Longer topics give better lyrics: names, places, a small story and the mood. Example: “my mom's 60th birthday, her garden, the songs she sang to us, grateful and happy”.

Lyrics Generated Results

Your lyrics will appear here

Give a topic on the left and click Generate Lyrics.
Vocal Isolator Input
Vocal Isolator Generated Audios

Your isolated vocals will appear here

Paste a track on the left and click Isolate Vocals.
Voice Cover Input

Upload a song with a clear lead vocal. It re-sings the lyrics in the new voice you pick, in the same style.

Voice Cover Generated Audios

Your voice cover will appear here

Fill in the form on the left and click Generate Cover.

Stem Splitter: Separate Vocals and Instrumental

Upload a song and Speechma Pro splits it into two stems: the vocals and the music without vocals. Both tracks are ready to play and download. Nothing to install.

Speechma Pro

Every AI voice tool, 50 million characters a month

No ads and no security codes. $4.89 $4.40 a month, cancel any time. See what you get

In short: Speechma Pro’s stem splitter separates a song into two tracks: the isolated vocals and a music-only (instrumental) track, and you can download both. Open Music Studio, choose Vocal Isolator, upload an audio file (up to about 9 minutes) or a short video, and press Isolate Vocals. Results are best on pop, acoustic and spoken tracks with a clear lead voice; on dense mixes a little music can remain in the vocal stem.

How to split a song into vocals and music

  1. Open the Vocal IsolatorIt is in Music Studio. On this page it is already open at the top.
  2. Add your songUpload an audio file (MP3, WAV, M4A and more) of up to about 9 minutes, or an MP4 or MOV clip. You can also paste a link to the file.
  3. Press Isolate VocalsPress Isolate Vocals. The AI pulls the voice away from the instruments.
  4. Get both stemsTwo results appear: Isolated vocals and Music only (no vocals). Play each one to check it.
  5. DownloadSave the stem you need, or both, and drop them into your editor, DAW or video project.

What each stem is good for

StemWhat you hearPopular uses
Isolated vocalsThe lead and backing voices with the band removedRemixes, a cappella edits, learning harmonies, clearer lyric transcription
Music only (no vocals)The full arrangement with the singing removedKaraoke, practice tracks, covers, background music for your own voiceover

Speechma Pro splits a song into two stems. It does not separate drums, bass and other instruments into their own tracks.

How clean is the split? A honest answer

AI stem splitting works from the finished stereo mix, not the original studio tracks, so it is a very good estimate rather than a perfect copy. Here is what to expect:

  • Best results: one clear lead singer, acoustic or pop arrangements, spoken word over music, and songs with space around the voice.
  • Harder cases: dense rock or metal mixes, heavy reverb, distorted vocals, live recordings with crowd noise, and choirs. Here some music can bleed into the vocal stem, and you may hear faint traces of the singer in the instrumental.
  • Quality in, quality out: a WAV or high-bitrate MP3 splits more cleanly than a low-quality file pulled from a phone recording.

Tip: listen to the loudest chorus and the quietest verse of each stem before you use it. Those two spots show you whether the split is good enough for your project.

Ideas for your stems

Karaoke nights

Use the music-only stem as a sing-along track. Our karaoke track maker page has more tips.

Practice and covers

Play or sing along to the instrumental, or learn a vocal line by looping the isolated vocals.

Remixes and mashups

Put an a cappella over a new beat, using songs you have permission to remix.

New voice, same song

Re-sing a track in a new voice style with AI Voice Cover.

Lyrics transcription

Run the vocal stem through Speech to Text for more accurate words.

Cleaner video audio

Pull speech out from under loud music in a clip you filmed yourself.

Use music you have the right to use

Splitting a song for personal practice is one thing; publishing it is another. Only share stems, remixes or covers of songs you own, songs that are licensed for it, or songs whose rights holder gave you permission. Need music you can as you like publish? Make your own with the AI music generator or AI song generator.

Frequently asked questions

Is there a stem splitter online?

Yes. Speechma Pro’s Vocal Isolator splits a song into a vocal stem and a music-only stem, with nothing to install.

Can I download both the vocals and the instrumental?

Yes. After the split you get two results, Isolated vocals and Music only (no vocals), and both can be downloaded.

How long can the song be?

Audio files of up to about 9 minutes work. You can also upload an MP4 or MOV clip.

Can it separate drums, bass and guitar?

No. It makes two stems: vocals and everything else. For karaoke, covers and a cappella edits, those are the two you usually need.

Why can I still hear some music in the vocal track?

On dense mixes, heavy reverb or live recordings, some music can bleed into the vocals. A higher-quality source file and a cleaner mix give better results.

Can I use the stems on YouTube or Spotify?

Only if you have the rights to the original song. The tool is included in Speechma Pro, but it does not change who owns the music.
The Speechma newsletter

Make better voiceovers, faster

  • New voices and features the day they launch
  • Script templates and tricks that make AI voices sound human
  • Creator stories, community picks and early access

About twice a month, one click to leave. We never share your email. Privacy · Read past issues

Voice Effects

0
0
100%