AI music generator

How to Make AI Music with ChatGPT and ElevenLabs

You can make music with AI by asking ChatGPT for lyrics, then entering those lyrics in ElevenLabs to generate a song. In my example, that process produced two variations of a 1990s Memphis rap track without uploading my voice.

The song came with generated vocals, music, and a title: “Midnight in Memphis.” The process starts with choosing a musical direction instead of asking the tool for something random.

Choose a Song Style and Generate Lyrics

For this demonstration, I chose 1990s Memphis rap. Rather than immediately pressing Generate in the music tool, I opened ChatGPT and asked it to write lyrics for that style.

The resulting lyrics established a clear setting: North Memphis at midnight. They described a black Chevy cruising through the neighborhood, bass playing, fog over the pavement, and streetlights overhead.

Those details became the starting point for the music generation. ChatGPT supplied the words, and ElevenLabs handled the musical performance.

The title “Midnight in Memphis” appeared with the generated results, rather than being a title I needed to supply at the beginning.

Generate the Song in ElevenLabs

Open the ElevenLabs AI music generator to follow the same workflow. ElevenLabs supports generating songs with vocals from text prompts, which is the capability used in this demonstration.

The dashboard steps are straightforward:

  1. Open the ElevenLabs dashboard and select Home.
  2. Choose Music from the available boxes.
  3. Enter the lyrics generated in ChatGPT.
  4. Generate the music and listen to the resulting variations.

The music interface also allows you to describe the musical style, instrumentation, and vocal style. In this example, the lyrics themselves provided cues about the kind of track I wanted.

Generations use credits. Each time you generate music, ElevenLabs deducts credits from your account. The demonstration produced two versions of the song.

Listen to Both Versions of “Midnight in Memphis”

Both versions used the same core lyrical ideas, but they offered different performances of the track. The spoken openings referenced Tennessee, North Memphis, and 1995 before moving into the verses.

The lyrics focused on a late-night drive through the neighborhood. Both tracks paired that setting with a bass-heavy, old-school rap sound.

Playing the versions back to back made the two interpretations easy to hear. The words and setting stayed familiar, while the delivery and musical treatment differed.

I supplied lyrics from ChatGPT without uploading my voice or asking ElevenLabs to imitate another person’s voice.

That distinction matters to what this example demonstrates. The generated vocals didn’t require a recording of me singing or rapping.

I also acknowledge in the video that some listeners don’t want to listen to AI-generated music. The purpose here is to demonstrate what happens when lyrics from one AI tool become the input for another.

Create a YouTube Title and Thumbnail Text

The bonus step uses our Free AI YouTube Title Generator to create publishing ideas for the track.

I entered “Midnight in Memphis” as the topic, then added more detail about the song. The example described a 1990s Memphis rap track, identified rap listeners as the audience, used a storytelling style, and selected an exciting tone.

After selecting Generate Titles, the tool returned options such as:

“Midnight in Memphis: A 90s Memphis Rap Story”

It also supplied thumbnail text suggestions. That means the same song concept provided the starting point for both the YouTube title and the words that could appear on its thumbnail.

Our generator provides wording you can review and edit. Those suggestions help describe the video, but they don’t promise views or rankings.

Turn a Song Idea Into Generated Music

The main result was a complete musical interpretation of lyrics written in ChatGPT. ElevenLabs supplied two versions of “Midnight in Memphis,” including vocals and a generated title.

The creative direction began with a specific style: 1990s Memphis rap. That choice carried through the lyrics, the sound of the generated tracks, and the YouTube title suggestions.