In short: To make an AI song, write or generate lyrics with section tags like [Verse 1] and [Chorus], then describe the sound in one line: vocal type first, then genre, tempo, mood and instruments (for example “female vocal, country pop, 95 bpm, warm, acoustic guitar”). On Speechma Pro, pick a length from 30 seconds to 8 minutes and one of 50 sung languages, then press Generate Song. It is included in Speechma Pro, and commercial use is included.
How to make an AI song in 6 steps
- Get your lyricsWrite your own, or open the Lyrics tool, type a topic, choose one of 50 languages and “Short (verse + chorus)” or “Full song”. Edit the result so it sounds like you.
- Choose the lyrics modeIn the Song tool, pick Write my lyrics and paste them, or Auto lyrics from a topic and type something like “a road trip from Austin to the Gulf Coast”.
- Write the musical styleOne line, vocal type first:
male vocal, 90s rock, 120 bpm, energetic, distorted guitars, live drums. - Set length and languageChoose 30 to 480 seconds (8 minutes) and the sung language, or leave it on Auto-detect from lyrics.
- Add options (optional)Under More options, tick Instrumental only (no vocals) for a backing track, or add a Style reference track whose sound the song should borrow.
- Generate and listenPress Generate Song, listen on headphones and download. Not right? Change one thing and try again.
How to write lyrics the AI can sing
Put each section tag in square brackets on its own line. Tags tell the AI where the verse ends and the chorus starts:
[Intro] [Verse 1] Porch light on and the screen door hums Waiting on the sound when your truck comes [Chorus] Come on home, come on home tonight Leave the highway to the neon light [Verse 2] ... [Bridge] ... [Outro]
- Common tags:
[Intro],[Verse 1],[Pre-Chorus],[Chorus],[Bridge],[Outro]. - Short lines sing better. Aim for 6 to 10 syllables per line.
- Write the chorus out every time you want it repeated.
- Spell words the way they are sung: “gonna”, “’cause”, numbers as words.
- Keep the mood consistent: sad lyrics plus a party style confuses the result.
Style prompt recipes you can copy
Formula: vocal type + genre + tempo + mood + 2 or 3 instruments. Do not name real artists; describe the sound instead.
| Use | Style prompt |
|---|---|
| Birthday song | female vocal, upbeat pop, 115 bpm, joyful, ukulele, hand claps, glockenspiel |
| Love song or first dance | soft male vocal, acoustic ballad, 72 bpm, tender, fingerpicked guitar, light strings |
| Country story | female vocal, modern country, 95 bpm, warm, acoustic guitar, pedal steel, brushed drums |
| Hype track | male rap vocal, trap, 140 bpm, confident, 808 bass, hi-hats, dark synth |
| Shop jingle | bright female vocal, retro pop, 120 bpm, cheerful, electric piano, bass, claps |
| Worship or choir | choir vocals, gospel, 80 bpm, uplifting, Hammond organ, piano |
| Indie rock | male vocal, indie rock, 128 bpm, nostalgic, jangly guitars, live drums |
How long should your song be?
| Length setting | Fits | Good for |
|---|---|---|
| 30 to 60 seconds | One verse and one chorus | Reels, TikTok, jingles, intros |
| 90 to 150 seconds | Two verses, two choruses | Birthday songs, gifts, demos |
| 150 to 240 seconds | Two or three verses, bridge, final chorus | A full radio-length song |
| Up to 8 minutes | Long lyrics or extended instrumental parts | Story songs, mixes, background music |
Match length to your lyrics. Very long lyrics in a short song get rushed; short lyrics in a long song leave long instrumental gaps.
Troubleshooting common AI song problems
- Sounds instrumental, no singing: name the vocal type at the start of the style, such as “female vocal” or “male vocal”, and check that Instrumental only is not ticked.
- Words are mumbled or skipped: shorten lines, slow the tempo (try 80 to 100 bpm), and fewer instruments help the voice cut through.
- Wrong language or accent: set the sung language yourself instead of Auto-detect.
- Chorus does not stand out: repeat the full chorus text each time and add “catchy chorus” or “big chorus” to the style.
- Song ends too early or drags: adjust the length to fit the amount of lyrics (see the table above).
- Not the sound you imagined: add a short Style reference track of music you have the right to use, or name two or three specific instruments.
- Mood feels off: make the lyrics and the style agree, for example “melancholy” lyrics with “slow, minor key” style.
Sharing your AI song
Download the song and share it anywhere: send it to family, post it on Instagram or TikTok, or use it in a YouTube video or ad. Speechma Pro songs include commercial use, so they work for client projects and businesses too. On YouTube, use the AI use setting in YouTube Studio when AI-generated music is the main focus of the video; YouTube says disclosure does not affect a video’s eligibility to earn money.
Use your own lyrics or lyrics you have rights to, and do not upload someone else’s song as a reference to copy it. To compare other tools, see the best AI song generators.
Frequently asked questions
How do I make an AI song?
Open Speechma Pro’s Song tool, paste your lyrics or choose auto lyrics from a topic, describe the style, pick a length and language, and press Generate Song. It is included in Speechma Pro.
What should I write in the style box?
Start with the vocal type, then genre, tempo, mood and two or three instruments. Example: “female vocal, upbeat pop, 115 bpm, joyful, ukulele, hand claps”.
Why does my AI song have no vocals?
Usually because the style does not name a vocal type, or Instrumental only is ticked. Put “female vocal” or “male vocal” first in the style.
What languages can an AI song be sung in?
Speechma Pro’s Song tool sings in 50 languages, including English, Spanish, Hindi, Portuguese, French, Korean, Japanese, Chinese, Arabic and Tagalog. The Lyrics tool writes in the same 50 languages.
How long can an AI song be?
From 30 seconds to 8 minutes on Speechma Pro. About 2 to 3 minutes fits two verses, two choruses and a bridge.
Can I use my AI song on YouTube or in ads?
Yes. Speechma Pro songs include commercial use. On YouTube, disclose AI-generated music that is the main focus of a video using the AI use setting.
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