What the AI labels mean
Everything on LISTN AI is made with AI. Two labels tell you how: the tool the creator says they used, and a score from our AI detector. They answer different questions, so they do not always agree.
Made with
The AI tool the creator says the song, episode or book was made with. Every upload must name one, and naming the wrong tool breaks our rules. It is the creator's own statement.
AI detector score
Our own check on the audio of a music track. It says how likely the whole song is to come from an AI music generator, from 0% to 100%.
Reading the AI detector score
The detector is confident the song comes from an AI music generator it knows.
It leans towards AI made, but is not sure. Heavy mastering or effects can blur the signs.
It did not find the fingerprint it knows. That does not mean a person made it.
It is a likelihood, not an amount
A score of 5.2% does not mean 5.2% of the song is AI. It means the detector found only a faint trace of the patterns it looks for. On the track page you may also see how many of the song's parts (about 30 seconds each) carry the fingerprint.
Why an AI song can score low
The detector was trained on songs from Suno (up to version 5) and Udio (up to 1.5). Music generators build sound with a neural vocoder that leaves small, regular peaks in the 1 to 8 kHz range; each generator leaves its own. Songs from tools it has not learned, such as YuE2, ElevenLabs or MusicGPT, can score low even though they are fully AI made. So a track labelled "Made with YuE2" with a 5% score is not a contradiction.
Where the score comes from
Each music track is checked when it is uploaded, and our team can check it again. When the detector gets a track wrong, our team can set the right percentage by hand. The detector is the open source AI Music Detector (MIT licence). Podcasts and audiobooks have no score: it only knows music.
Human touch and How it was made
Human touch means the creator made part of it by hand: wrote the lyrics, sang, played an instrument, mixed or mastered it, wrote the script or narrated. The "How it was made" box on each release, show and book says which parts, the tool's version, the prompt when the creator chooses to share it, and the creator's own notes.
It is the creator's own statement, like "Made with". Our team checks releases before they go live, and a false statement breaks our rules.
The AI tools
Hover over a "Made with" label anywhere on LISTN AI to see this. Press one to hear everything made with that tool.
An AI music generator that writes complete songs, with vocals and lyrics, from a short text description.
The AI detector knows this tool.An AI music generator that makes songs with vocals from a text description, and can extend and remix them.
The AI detector knows this tool.Stability AI's model for music and sound from text, best known for instrumentals and loops.
An AI composer for instrumental and orchestral music, often used for film, game and background scores.
Makes instrumental tracks from a mood, a genre or a few tags.
Makes a complete track in a chosen style in a few clicks, mostly instrumental.
Makes instrumental music you shape by mood, length and structure.
Composes instrumental background music to fit a mood or a video.
An AI music generator for instrumental tracks in many genres.
Makes music, including vocals, from a text description.
Meta's open source model that makes instrumental music from a text description.
Google DeepMind's music model, which makes music with or without vocals from text.
An open source model that writes full songs with vocals from lyrics, running on the creator's own computer.
An AI service that makes songs, vocals and instrumentals from a text description.
AI voices for narration, hosts and characters, and AI music. Used here for podcasts, audiobooks and songs.
OpenAI's text to speech voices, used to read scripts and books aloud.
Turns documents into a conversation between two AI hosts (Audio Overviews).
AI voiceovers for narration and explainers.
Studio style AI voices for narration.
AI voices for narration and podcasts.
An audio and video editor with AI voices that read a script.
Text to speech with natural AI voices.