HeartMuLa

HeartMuLa is an AI song generation tool consisting of two parts: first, an online web application that can turn text descriptions or your own lyrics into complete songs with vocals in about 1–2 minutes; second, the open-source music model HeartMuLa (Apache 2.0 license, using the 12.5 Hz HeartCodec, with song lengths of up to about 6 minutes), which you can deploy and run yourself.

Main uses

HeartMuLa is aimed at users who want to quickly create original music without relying on the traditional arrangement process. The officially described use cases include: background music for videos (YouTube videos, short films, social media content), podcast intros and outros, soundtracks for social media such as TikTok/Instagram Reels/YouTube Shorts, indie game soundtracks, and personalized song gifts for special occasions such as birthdays and weddings; the Pro and Business plans also allow commercial use for advertising, product videos, and brand content.

Main capabilities

• Text to music: describe the desired mood, style, and genre in natural language, for example "an upbeat electronic dance track with euphoric synths and driving bass", and the AI generates music accordingly.
• Supports your own lyrics: you can write lyrics using structural tags such as [Verse], [Chorus], [Bridge], [Outro], and the model understands the song structure and generates vocals that match the lyrics. If no lyrics are provided, the AI can also help generate lyrics and the complete song structure.
• Style tag control: provides hundreds of style tags (such as ambient, cinematic, lo-fi hip hop, orchestral) for fine-tuning the sound.
• Complete song structure: generates complete songs containing Verse, Chorus, Bridge, and Outro, and each generation produces two versions to choose from.
• Generation and export: each song takes about 1–2 minutes to generate, producing unique audio with vocals, instruments, and mastering. It is free to preview and share; after any purchase you can download MP3s, and the Pro and Business plans include commercial usage rights.
• Supported genres: includes Pop, Rock, Electronic, Hip-Hop, R&B, Jazz, Classical, Ambient, Lo-Fi, Country, Metal, Reggae, Latin, K-Pop, J-Pop, and more, and genres can be combined.

How to use

1. Sign up for free to get credits and create your first AI song; daily check-ins also earn you another song each week.
2. Describe your idea: write a prompt in natural language, specifying the desired mood, genre, and style.
3. (Optional) Add lyrics: write your own lyrics using structural tags.
4. Choose style tags and fine-tune the sound.
5. Click generate, and the model produces a complete song in about 1–2 minutes, generating two versions each time.
6. Listen and share for free; download MP3 after any purchase. Pro and Business plans support commercial use. You can also download the HeartMuLa 3B model from Hugging Face for local deployment, under the Apache 2.0 license, available for commercial use.

Comparison with Suno (official statement)

The official comparison page shows: Suno is a closed-source cloud service; the HeartMuLa web app also runs in the cloud, but unused credits do not reset monthly, and there is no monthly cap on downloads after any purchase. In addition, HeartMuLa also offers an open-source model (Apache 2.0) that you can run yourself. The official note states that the Suno-related information comes from suno.com/pricing, checked in October 2026.

Pricing and licensing (official statement)

• Free start: sign up to get one free song, and daily check-ins earn another song each week; after that, you can choose a subscription plan or a one-time credit pack.
• Plans: Pro and Business plans allow commercial use; one-time credit packs require no subscription; Top-ups need a plan.
• Commercial scope: songs made during a Pro or Business subscription can be used commercially (such as YouTube monetization, podcasts, ads, games, releases, and client projects), with no attribution required, and the license remains valid after canceling the subscription. Songs made with the Free or Starter plans or credit packs are for personal use only. Self-run open-source models are subject to their Apache 2.0 license. (Specific price figures are not fully provided in the official text and are not yet available.)

System requirements (for the open-source model)

The HeartMuLa 3B model requires about 24GB of VRAM. Recommended GPUs include RTX 3090, RTX 4090, or A100. Users with insufficient VRAM can use cloud GPU services such as RunPod and Vast.ai.

Limitations and notes

• The web app is a cloud-hosted service and requires registration; free quota is limited.
• Capabilities such as MP3 downloads and commercial use are tied to specific plans, and the specific prices and quotas are not fully given in the official text.
• The open-source model requires hardware such as 24GB or more of VRAM, so the configuration threshold is relatively high.
• AI music differs from stock music: each request is original, will not be the same as others', and you can specify mood, genre, lyrics, and duration.