A better SoundBreak song starts with a clear brief and the right writing mode. In a Quick Session, your prompt starts an audio generation. It does not return written lyric options, compare chorus drafts, or explain its choices. In a Guided Session, you shape the song step by step, including its theme and lyrics, before generating audio.
The lyric-craft advice is the same in both modes: give the song a specific job, protect the hook, use concrete images, and revise one problem at a time. The product steps are different, so choose the mode before you decide how much control you need.
Use Quick Session when you have a clear brief and want to hear it quickly. Choose an AI Cowriter or On your own, then write one prompt. A story-only prompt keeps an AI Cowriter's genre, vocals, and instruments. A prompt can override vocal gender, vocal style, instruments, tempo, and era.
Quick Session includes these tools:
The write box generates audio. Do not paste a request such as “return three chorus options,” “compare these candidates,” or “explain the choices.” Those instructions do not open a lyric editor. They start a song generation, and the response is audio.
Guided Session is the paid step-by-step flow. Use it when you need to shape the brief and lyric before you hear the songs.
The flow asks you to:
[Verse 1] and [Chorus].This is the place to revise one section while keeping the hook. For example, tighten the chorus, keep the exact title phrase, or change the verse without replacing the whole lyric. In Guided Session, the vocal feel comes from the selected style clip, plus the Rap / Sing switch.
After the songs are generated, choose Reuse this setup to work from the same choices, or Start from scratch to begin again.
The picker label is On your own. Its subtitle is Your Style / Your Songwriting. Choose it when you do not want an AI Cowriter's default songwriting style.
The most useful brief answers one question first: what should this song or section make clear?
Examples:
“Make a catchy song” gives the model a direction but no useful test. “Write an intimate alt-pop song about leaving after one last train ride, with the hook ‘leave the light on,’ and a chorus that turns quiet restraint into resolve” gives you something to judge in the audio.
Start with the song job, then add only the details that help the generation solve it.
Quick Session prompts control the audio brief. They do not lock lyric text unless you put the exact lyrics in Add your lyrics.
Give the lyric a situation that can produce concrete lines.
A songwriter leaves an apartment before sunrise after deciding not to return.
Specific situations give the lyric somewhere to stand. A broad topic such as “heartbreak” can still work, but it needs a scene, memory, or decision attached to it.
Say who is speaking, who they are addressing, and what changes between sections.
First person, speaking directly to an ex-partner. The verse is tired and uncertain; the chorus turns that uncertainty into a calm decision.
Point of view helps the lyric avoid floating between “I,” “you,” and “we” without a reason. A section job keeps the chorus from simply repeating the verse's mood.
If you want an exact lyric, put it in Add your lyrics in Quick Session. Lyrics mentioned only in the prompt are direction, not locked text.
If you want the model to use a hook idea without supplying a fixed lyric, say what the hook should do:
Build the chorus around the idea of leaving the light on as a final sign of care.
If you need the exact phrase “leave the light on” to remain unchanged, add the lyrics rather than relying on the prompt.
With an AI Cowriter, a story-only prompt keeps that Cowriter's genre, vocals, and instruments. A prompt can override:
For example:
Midtempo alternative pop, lower-register lead vocal, clear and direct delivery, warm bass, muted electric guitar, around 100 BPM, current production with a late-1990s alternative-pop frame.
The Quick Session prompt cannot reliably set close-mic distance, breathiness, backing-vocal placement, or a verse-versus-chorus arrangement. Do not present those as Quick Session controls. If those details are central, use Guided Session with a style clip and the Rap / Sing switch, then judge the generated audio.
Constraints can guide the generation without pretending to guarantee the result.
Keep the lyric specific and uncluttered. Avoid motivational slogans, abstract filler, and a second plot twist. Make the chorus state the central decision clearly.
An avoid list is useful when the first result reaches for familiar phrases:
The following is an original text exercise. It demonstrates how to shape a lyric brief. It does not claim that any generated melody, vocal performance, or mix will work.
[Chorus]
I know I have to let you go
But I still see you everywhere
I know I have to let you go
I wish that you were still hereThe idea is clear, but the section has no specific scene, no memorable image, and no clear emotional turn. It repeats the conclusion without showing what caused it.
Before generating, decide what the chorus must do:
If you need to develop or revise the lyric itself, use the Guided Session lyrics step. Choose a first draft, work section by section, or upload a labeled sheet with tags such as [Verse 1] and [Chorus]. If you already have the exact lyric, use Add your lyrics in Quick Session.
[Chorus]
Cold keys on the counter
So leave the light on
The platform turns the windows blue
I walk before the dawnThis revision keeps the hook and gives the singer a body, place, and time. Read it aloud before moving to audio. If one line feels crowded, revise that line rather than replacing the whole lyric.
In Quick Session, the prompt can now describe the song around the lyric:
Midtempo alternative pop about leaving an apartment before sunrise after
choosing not to return to a relationship. First-person, restrained and
honest rather than inspirational. Lower-register lead vocal with clear,
direct delivery. Warm bass and muted electric guitar. Keep the lyric
specific and leave space for the hook.Put the exact chorus in Add your lyrics. Do not add a line asking SoundBreak to return lyric options or explain the result. The write box will generate audio.
If you need to change the chorus after hearing the song, make that change in the Guided Session lyrics step or edit your lyrics before starting the next Quick Session. The audio result is not a written candidate list.
Guided Session is the product path for the craft workflow above.
Guided Session does not turn the write box into a text-answer tool. The lyrics are shaped in the Guided lyrics step, and generation still produces songs.
In Guided Session, the selected clip is the sound of the final song. The clip is not a menu of traits that you can separately text-steer. Choose the clip for the overall vocal and musical feel you want, then use the Rap / Sing switch to choose the broad vocal mode.
In Quick Session, Upload song for style is paid only. It accepts MP3, MP4, or M4A files up to 10 MB. The selected clip must be at least 30 seconds, and it must be a song you created as the artist or creator. The attached file overrides style in the prompt.
Do not upload a purchased, downloaded, licensed, or other third-party recording as a style reference, even if you own a copy. SoundBreak's Acceptable Use Policy prohibits using material you do not have the legal right to use.
Instrumental is controlled by the Instrumental only switch in Quick Session or by the first Guided Session question. Do not attach lyrics and turn Instrumental only on. If you want a vocal song, leave Instrumental only off and provide lyrics in the correct lyrics field when you need exact text.
After a Quick Session song, reuse is organized around:
There is no tempo knob. To change tempo, say so in the next prompt:
Keep the same lyric and music style, but make the next generation slower and more spacious.
Change one thing at a time when the result is close. This makes it easier to tell whether the issue was the song idea, lyric, style, or tempo instruction.
Useful next prompts include:
These prompts start another audio generation. They do not ask SoundBreak to return a written comparison.
An AI Cowriter guides songwriting style. It is not that artist's voice, and the result is not the artist's performance.
You can use the Cowriter's default genre, vocals, and instruments with a story-only prompt, or override vocal gender, vocal style, instruments, tempo, and era in the prompt. Do not ask for a real artist's exact voice, and do not describe a generated track as a human artist's recording, endorsement, authorship, or identity.
If you choose On your own, SoundBreak starts from Your Style / Your Songwriting instead of an AI Cowriter's default style.
Hook. Can you find the memorable phrase by the second line? If it must remain exact, put it in Add your lyrics or the Guided lyrics step.
Point of view. Does the speaker stay consistent? Name the speaker, listener, and grammatical point of view.
Section job. Does the chorus change the verse's emotional state? Describe the turn in plain language.
Specificity. Is there at least one image that belongs to this song? Add two or three concrete objects, places, or actions.
Singability. Can you read each line without rushing? Shorten the line, remove a second idea, or revise that section in Guided Session.
Rights. Did you supply only lyrics and audio you created yourself? Replace anything you did not create.
Use this as an audio-generation brief. Put exact lyrics in Add your lyrics, not only in the prompt.
Create a [genre] song about [story, scene, or question].
Song job:
[What should the song make clear, change, or reveal?]
Point of view:
[Who is singing, who are they addressing, and what tense should the lyric use?]
Vocal direction:
[Vocal gender and vocal style, if needed]
Music:
[Instruments, tempo direction, and era or broad production frame]
Lyric direction:
[Hook idea, concrete images, and what to avoid]
Use the exact lyrics I attach if I provide them.This prompt starts an audio generation. It does not request lyric options, a comparison, or an explanation.
What should I include in a Quick Session prompt?
Include the song job, story or scene, point of view, vocal gender or vocal style if needed, instruments, tempo direction, era, and lyric direction. A story-only prompt keeps the selected AI Cowriter's genre, vocals, and instruments.
How do I lock exact lyrics?
Put lyrics in Add your lyrics in Quick Session or in the Guided Session lyrics step. Lyrics mentioned only in the prompt are not locked.
Where do I revise one chorus section and keep the hook?
Use Guided Session. Its lyrics step lets you work from a first draft, edit section by section, or upload a labeled lyric sheet. Keep the exact hook in the lyric you submit, then generate the revised song.
How do I prompt for stronger AI vocals?
In Quick Session, describe vocal gender and vocal style, then use instruments, tempo, and era to frame the song. Do not rely on the prompt to control close-mic distance, breathiness, backing-vocal placement, or verse-versus-chorus arrangement. In Guided Session, choose a style clip for the final song's sound and use the Rap / Sing switch.
Should I ask for two songs in the prompt?
No. Return 2 songs is a switch, on by default in Quick Session. It returns two takes of the same brief. Use the switch rather than adding a request for two directions to the prompt.
Can I use lyrics or audio from another artist?
No. Use only lyrics and audio you created yourself. The writing flow says uploaded style references must be songs you created as the artist or creator, and the Acceptable Use Policy prohibits using material you do not have the legal right to use.
Can I attach lyrics and turn on Instrumental only?
No. Instrumental is controlled by the Instrumental only switch or the first Guided Session question. Do not attach lyrics when you want an instrumental.
Can I change tempo with a knob after a Quick Session?
No. There is no tempo knob. Ask for a tempo change in the next prompt.
Can an AI Cowriter sound exactly like the artist?
No. An AI Cowriter guides songwriting style. It is not that artist's voice, and the result is not their performance.
Does a detailed prompt guarantee better audio?
No. A detailed brief gives you clearer choices to evaluate, but it does not guarantee a particular melody, vocal performance, or mix. Listen to the generated song and revise the lyric, setup, or next prompt based on what you hear.