HTML Table of Contents to blog topics to insert here
Introduction
I was recently introduced to Suno, and I have to say, I was almost instantly addicted!
I know little about music, except for what I like to listen to. However, I quickly realized my knowledge is incomplete. I don't know the names of many musical styles, which makes it a bit tricky to use Suno effectively.
I never thought I could create music before. Now, I find it so much fun to take simple ideas and see them implemented as full songs!
While many tutorials exist online, few offer a basic starting point for beginners like me. That's why I wanted to share my experience and some tips.
Cost
One thing that is annoying about Suno is that you can’t see the pricing on the homepage without signing up for an account first. The good news is, if you just want to dabble with it a bit, you can create 5 songs a day in pairs with 50 credits. These credits reset to 50 every night at what I'm guessing is midnight UTC. I'm not sure if daylight time will affect this, but I currently see my credits refresh at 7 PM CST.
When you sign up for a new account, you start with 100 credits that work with model v3.5 (limited to 3.5 minutes). You also get 20 credits as a one-time gift for model 4 (limited to 4 minutes). If you have long lyrics, your songs will just get cut off at those time limits.
Once you get into it and start really enjoying the process (let's be honest, "addicted" is a good word!), you'll want to make 10 or 20 songs in a row. One idea often leads to another, and having to wait a day would definitely interrupt the flow. That’s where a paid plan comes in handy.
They have three plans:- Free: 50 credits per day.
- Pro: $10 for 2500 credits per month, or $96 for a year (-20%) .
- Premium: $30 for 10,000 credits per month, or $288 for a year (-20%).
When you sign up for a new account, you start with 100 credits that work with model v3.5 (limited to 3.5 minutes). You also get 20 credits as a one-time gift for model 4 (limited to 4 minutes). If you have long lyrics, your songs will just get cut off at those time limits.
Once you get into it and start really enjoying the process (let's be honest, "addicted" is a good word!), you'll want to make 10 or 20 songs in a row. One idea often leads to another, and having to wait a day would definitely interrupt the flow. That’s where a paid plan comes in handy.
They have three plans:
- Free: 50 credits per day.
- Pro: $10 for 2500 credits per month, or $96 for a year (-20%) .
- Premium: $30 for 10,000 credits per month, or $288 for a year (-20%).
Premium looks like a better deal with 4x the songs for 3x the money, but the catch is that any unused credits at the end of the month don’t roll over.
Unless you’re a full-time musician, I think Pro is the better deal. You can also buy blocks of credits that don’t expire! They are sold in four block sizes at $4 per 500 credits, costing $4, $8, $16, or $30. That's 8 cents for each pair generated. If one used every monthly credit the cost/pair would be 4 cents on Pro, 3 cents on Premium, paying per month. If you use all your credits before a month is out, you'll get 50 a day like free accounts until the next month.
Here’s what Suno says about credits: “Credits included in subscriptions do not carry over from day to day or month to month. Purchased top-up credits do not expire, but require an active subscription to use.”
You can also earn credits by encouraging your friends to sign up and make 10 songs. If you don’t have a referral link from your own friends, please use mine:
Riffusion:
After playing with Suno for a while, I ran accross this video about Riffusion:
A New AI Music King is Here and it's Totally FREE & Unlimited!
https://www.youtube.com/watch?v=uBmLqOTs2jo
https://invidious.f5.si/watch?v=uBmLqOTs2jo
Please use my referral link to try it!
https://riffusion.com?r=WhiteMagic
Riffusion Caveats:
Studio mode burns an hour when turned on, so don't try to get more time by turning it on and off between generations as that will just burn it up much faster!
Song length is limited to 4 minutes like the free models on Suno, but lyrics are often not performed as song approaches a 4 minute length.
Suno Likes:
Great social media features with promoted genres, playlists, hashtag challenges, and suggested similar songs.
Left menu links can all be opened in new tabs.
Songs generally have one decent rendition on the first try.
New features come out every few weeks!
Feature suggestion sections in Discord.
Suno Dislikes:
No light theme available, and the CSS is too complex to easily roll your own!
There is no unlimited mode, pricing is complex and not a bargain.
Due to rapid feature development, not all bugs are addressed as quickly.
Limited Direct External Sharing & No Embedding: While you can share links to your Suno songs on external social media, there isn't a direct, seamless integration for sharing like embedding playable audio directly into a tweet or an Instagram story. Crucially, Suno actively blocks embedding via iframes, meaning you cannot easily host a playable version of your Suno song directly on your own website or blog without directing users back to the Suno platform.
Riffusion Likes:
Unlimited song generation even at the free level (with modest 1-5 minute queue).
Decent quality results, but a seemingly narrower training base.
Clean and simple light theme with responsive CSS so it works nicely on any device without the need for a dedicated app.
Riffusion Dislikes:
While every menu item has its own link that you can bookmark, you can't right-click them to open in a new tab to compartmentalize projects.
Both Suno and Riffusion have active Discord groups.
HTML TABLE of Music Generation sites with a free level to be inserted here.
Ideas to expand:
Running similar open source software on your own computer:
Install YuE 7B Locally to Generate Full Songs with Music from Lyrics
https://invidious.reallyaweso.me/watch?v=RSMNH9GitbA
https://youtu.be/RSMNH9GitbA
llamusic/llamusic model to create lyrics with extensive style suggestions
https://ollama.com/llamusic/llamusic
suno-ai/bark: 🔊 Text-Prompted Generative Audio Model
https://github.com/suno-ai/bark
CoT vs. ICL Models:
Designed for generating songs primarily based on lyrics and genre prompts, a "Chain-of-Thought" (CoT) approach, meaning it tries to reason through the lyrical content and genre to produce a coherent and musically appropriate song from scratch.
In-Context Learning (ICL) means you can provide it with a reference audio segment (e.g., 30 seconds of an existing song, either instrumental, vocal, or mixed). The model then tries to generate a new song with new lyrics that has a similar style, vocal characteristics, or musicality to the provided audio reference.
Phonetic spellings will likely be pronounced better than unusual acronyms.
Suno Model version differences:
3.5 will repeat a single short verse
4.5 will play it once, then fill the rest of the time with instrumentals and maybe some vocalizations w/o words. If using a persona, garbled lyrics from the original song can (often?) take over!
Personas can introduce words as a mix of the original song. After loading the persona, delete it again just keeping the styles to include and exclude.
simple prompts & hints, parody & satire auto lyrics to approximate favorite songs as artist names are not allowed in styles,
Shorthand to squeeze examples into 200 character limit prompts, | instead of or, & instead of and, omit spaces like "Parody:SoundofMusic",
Use examples to clarify a theme
highlight best songs with better graphics as Suno's art is convenient, but usually not eye catching
Browse other folks songs for inspiration, examples. Train your inner AI! ;-)
Thumbs up & Follows will bring your own songs attention as well. Use playlists to track examples you like. This will also help you find more songs of the styles you create or play most on the "For You" list:
https://suno.com/for-you
which is part of the many lists on the Home page.
Follow rather than just like other folk's personas so you can easily find them in your own library.
Hack the URLs for playlists of your own custom genres not on the master list. e.g.
Happiness is an inside jobFeatures I'd like to have like add multiple songs to a play list, or select multiple playlists for one song (available in the Android app)
A listen later button (Well add to queue may do?)
A repeat button for a playlist or track (like in VLC)
Automations like bulk song download to backup your work. Site terms allow them to delete anything deemed offensive!
Reference links like legal analysis:
Suno Exposed: Terms Of Service...
(Don't Release AI Music Until You Watch)
https://invidious.perennialte.ch/watch?v=HlGIxLH1K-M
https://youtu.be/HlGIxLH1K-M
Riffusion Exposed: Terms Of Service... Watch Before You Sign!
https://invidious.reallyaweso.me/watch?v=cDjaufNY4tY
https://youtu.be/cDjaufNY4tY
Suno prompts - Prompting AI music (Warning several ads for paid guides)
https://howtopromptsuno.com/
Note: Music Generation sites all pretty much own any lyrics you generate from a prompt. Save you credits and use an AI like Gemini or Copilot to write your lyrics so you at least own that portion of a song that might get popular!
Free AI tools for better images and video loops:
AI Video Generator: Text & Image to Stunning Video, Simple & Quick | Hailuo AI
https://hailuoai.video/create
Free AI Video Generator - Bing Video Creator [app only for now, several vids/day]
https://www.bing.com/images/create?ctype=video
LightX [One free video clip per day]
https://www.lightxeditor.com/ai-photo-editor/edit?tab=text2video
GitHub - hpcaitech/Open-Sora: Open-Sora:
Democratizing Efficient Video Production for All
https://github.com/hpcaitech/Open-Sora
GitHub - THUDM/CogVideo: text and image to video generation: CogVideoX (2024) and CogVideo (ICLR 2023)
https://github.com/THUDM/CogVideo
Song Examples:
lyrics likely from simple prompts
Catchy graphics
Useful Directives
creative styles
Hacking AI Music sites with UserScripts:
https://greasyfork.org/en/scripts?q=suno
https://greasyfork.org/en/scripts?q=riffusion
Related Articles:
AI is coming for music, too
https://www.technologyreview.com/2025/04/16/1114433/ai-artificial-intelligence-music-diffusion-creativity-songs-writer/
References:
YuE-s1-7B-anneal-en-cot | AI Model Details
https://www.aimodels.fyi/models/huggingFace/yue-s1-7b-anneal-en-cot-m-a-p
https://github.com/alisson-anjos/YuE-Interface/blob/main/README.md#model-suffixes-explained
No comments:
Post a Comment