iLoveSong.ai verdict
iLoveSong.ai is the clearest choice for Brazilian Portuguese creators who want to go from a lyric idea to a shareable MP4 singer video in one tool, with Google Lyria 3 access included on every paid plan. The annual plan at $7.30/month is genuinely affordable, but there is no free tier at all, which puts it behind Suno and Udio for anyone who wants to test before committing. If you need professional voice cloning or text-to-speech rather than song generation, ElevenLabs is the better fit.
Strengths
Google Lyria 3 is included on every paid plan, not gated behind a premium tier, giving access to a high-quality generation model at the $7.30/month annual price.
The AI Singer Video Generator produces a lip-synced MP4 from a single portrait photo, removing the need for a separate video tool for short-form content.
Weaknesses
No free tier exists at all: the cheapest entry point is $14.80/month billed monthly, which is a hard barrier for anyone evaluating output quality before committing.
Commercial licensing is not included on the Basic monthly plan according to the pricing page, meaning the most accessible tier leaves commercial use rights ambiguous for creators who need them.
Neume verdict
Neume is the right pick for anyone who wants a finished, downloadable song fast and has no interest in learning music production controls. Its stem separation tops out at two stems and there are no BPM, key, or mode settings, so producers who need granular control will find it thin compared to tools like Suno or Udio. If you need multi-stem export or MIDI, look elsewhere.
Strengths
Commercial rights included from the first free song, with no paid-tier requirement to keep or sell what you generate.
Section-level Remix feature rewrites lyrics in a specific time range without regenerating the whole track, a genuinely useful editing shortcut not universally available in this category.
Weaknesses
Stem separation produces only two stems (vocals and instrumental), not the four-plus stems that producers need for mixing; there is also no MIDI or DAW export of any kind.
No fine-grained production controls: the interface has no BPM, key, mode toggle, or vocalist selector, so the model's output on those dimensions cannot be directly specified.
