AI Music Generator

Generate music and sound effects from a text prompt in AI Studio. Each model shows its credit price per run before you press Generate.

ElevenLabs Music

ElevenLabs documents tracks with vocals or instrumental only, in several languages including English, Spanish, German and Japanese. Source

ElevenLabs Music v2

2 credits per run at the default settings (the shortest or lightest run; longer or larger runs cost more). Input: text prompt.

  • Force Instrumental: on or off
  • Duration: 3 to 20 seconds

ElevenLabs Music v2.5

4 credits per run at the default settings (the shortest or lightest run; longer or larger runs cost more). Input: text prompt.

  • Force Instrumental: on or off
  • Duration: 3 to 20 seconds

Seed Audio

Seed Audio 1.0

1 credit per run at the default settings (the shortest or lightest run; longer or larger runs cost more). Input: text prompt, optional image, optional audio reference.

  • Format: wav, mp3, pcm, ogg_opus
  • Sample Rate: 8000, 16000, 24000, 32000, 44100, 48000

Sonilo Music

Sonilo Music

1 credit per run at the default settings (the shortest or lightest run; longer or larger runs cost more). Input: text prompt.

  • Duration: 1 to 20 seconds

Mirelo sound effects

Mirelo builds models that generate sound for video. Source

Mirelo Text to Audio

3 credits per run at the default settings (the shortest or lightest run; longer or larger runs cost more). Input: text prompt.

  • Duration: 1 to 20 seconds

Real runs by the Twistavo team

Results of runs the team made, with the exact prompt and settings. Files are shown as delivered unless the output line says otherwise. The price is the server quote for the same run today.

Input
A text prompt written by the team, no audio input.
Prompt
Show the full prompt

high-energy hype trap/EDM instrumental, hard 808 bass, building tension intro then a hard beat drop around 2-3 seconds, aggressive sports hype anthem, no vocals

Tool and model id
Sonilo Music (sonilo-music)
Settings
duration 10
Output
M4A (AAC), stereo, 44.1 kHz, 10 s, 367 KB (the file as delivered)
Price
6 credits for this run
Run date
2026-06-17
  • Integrated loudness is -13.7 LUFS with a loudness range of 2.1 LU (ffmpeg ebur128): a loud, mostly even level.
  • The level dips between about 2 and 3 s (from -9 to -21 dB RMS per half second) and comes back up at 3.5 s; the prompt asked for a drop around 2 to 3 seconds.
Input
A text prompt written by the team, no audio or image input.
Prompt
Show the full prompt

A 15-second scene audio: dark dramatic reveal, a deep tense pulse, dramatic low strings swelling, one strong deep percussion accent in the middle, oppressive verdict mood. Instrumental only, no speech, no words.

Tool and model id
Seed Audio 1.0 (seed-audio)
Settings
format wav, sampleRate 48000
Output
WAV, stereo, 48 kHz, 15 s, 2.7 MB (converted here to AAC at 128 kb/s, 240 KB, so the page stays light)
Price
1 credit for this run
Run date
2026-07-01
  • The clip is exactly 15.0 s long. Seed Audio has no duration setting: the length came from the prompt ("A 15-second scene audio").
  • It starts loud, stays near -30 dB RMS from about 2 to 8 s, then rises twice, near 9 s and near 13.5 s (to about -14 and -11 dB RMS). The prompt asked for one accent in the middle.
  • Integrated loudness is -21.3 LUFS with a loudness range of 11.9 LU: quieter and far less even than the Sonilo track.
Input
A text prompt written by the team, no audio input.
Prompt
Show the full prompt

Coffee being knocked over and spilling, liquid splashing onto a stack of paper documents on a wooden desk, crisp and clear

Tool and model id
Mirelo Text to Audio (mirelo-text-to-audio)
Settings
duration 2
Output
MP3, mono, 44.1 kHz, 2 s, 16 KB (the file as delivered)
Price
5 credits for this run
Run date
2026-06-26
  • Very quiet: integrated loudness is -42.6 LUFS. Raise the gain before you use it in an edit.
  • The level is under -45 dB for most of the clip: sound shows only in the first 0.08 s, around 0.75 to 0.82 s, and from 1.74 s to the end (ffmpeg silencedetect).

How to run it in the studio

  1. Open AI Studio and choose the Audio tab.
  2. Pick a model and give it what it takes: a description of the track or sound you want in the prompt box.
  3. Set the options, then press Generate: the button shows the credit price first.

Questions

How much does one run cost?

Each model above shows its credits per run at its default settings. In AI Studio the Generate button shows the exact price of your settings before the run starts.

How long can a track be?

Models with a Duration option run from 1 to 20 seconds; each model's own range is listed above, and longer tracks cost more credits.

Other tools