Creative ProductionAI ChatMusic & AudioPrompt Enhancement

Creators, marketers, musicians, and audio producers

Music & Audio Production

Develop an original song or instrumental brief, choose a compatible music workflow and review the returned track against its intended use. Keep songwriting instructions, lyrics, app controls and the audio you actually receive clearly separated.

Practical guidance

Know what you are making

A reusable music brief, route-appropriate generation input and listening checklist. A finished edit, exact duration, cleared usage rights or a particular sound quality is not guaranteed.

Before you start

Purpose and musical direction

Describe the intended use, mood, instruments, energy and whether vocals are wanted. A desired edit length belongs in the creative brief; it is not a duration selector. Avoid copying an existing song or requesting an identifiable performer's voice.

Original or authorized words

Prepare lyrics separately in AI Chat when appropriate. Confirm permission to use any supplied material. Remove private information and do not assume generation itself grants permission to release or advertise with the result.

A route-specific input plan

Read the chosen model's guide first. The owner audit reports a general 1–500 character prompt; Suno Custom mode has separate fields, while MiniMax reuses the main prompt as lyrics. Long planning prompts below belong in AI Chat, not the music prompt box.

Practical guidance

From input to reviewed output

  1. Write a compact creative brief

    Use AI Chat to name the track's purpose, mood, instrumentation, energy arc and exclusions. Choose one coherent direction rather than stacking incompatible genres.

    Checkpoint: Separate desired creative behavior from supported app controls and provider limits.

  2. Match the brief to a model route

    For Suno, distinguish Standard from Custom input. For MiniMax, plan the main input as words that may be sung. For ElevenLabs, understand the AIML route and reported fallback before relying on a particular model identity.

    Checkpoint: Verify the chosen option is available in your session; catalog presence alone does not prove current access.

  3. Prepare the actual generation input

    Shorten a descriptive brief to the applicable limit or prepare the correct lyric field. Do not paste instructions such as review this campaign or explain your reasoning into a field used as lyrics.

    Checkpoint: Count characters after replacing placeholders. Suno custom-lyrics serialization remains unverified; do not promise exact delivery of custom words.

  4. Generate only when ready, then listen

    If you choose to generate in the app, inspect the returned audio rather than judging by the prompt or model label. Record audible words, unwanted sounds, arrangement and actual length.

    Checkpoint: The templates here were not generation-tested. Do not repeatedly submit a pending job merely because it has not finished.

  5. Decide whether to revise or edit

    Change one musical property at a time for a new attempt. If an otherwise suitable track needs timing, fades or a mix under narration, plan those changes in an external audio/video editor unless the needed app capability is separately confirmed.

    Checkpoint: Review the final edited audio and applicable usage terms before publishing; retain the brief and your listening notes.

Technical details

What the workflow does—and does not do

Owner audit; output not generation-tested

Requested duration is not a user control

The October 9 owner-supplied audit reports durationSeconds: 60 and no user-selectable duration control. That request is not forwarded for Suno, MiniMax or the Lyria routes. It is not a guarantee of 60-second output; check the returned track. See individual guides for other duration mappings and unresolved parameter mismatches.

Official gateway contract + owner-audit boundary

Suno Custom has a distinct contract

aivideoapi documents a 500-character Standard prompt. Custom mode uses title up to 80 characters, style up to 200 and, for vocal songs, lyrics in input.prompt up to 3,000. The owner audit reports separate Custom fields, but exact upstream custom-lyrics serialization is unverified. Extra gateway controls are not automatically available in the app.

Owner audit contrasted with gateway documentation

MiniMax receives the main prompt as lyrics

The audit reports prompt = main prompt and lyrics = main prompt, with no separate custom-lyrics input. WaveSpeed Music 2.6 documents separate fields and an is_instrumental setting; the audited app does not forward that setting. Its lyric schema specifies 10–3,000 characters, unlike the app's general 1–500 allowance. Acceptance and output behavior were not tested.

Official gateway contract + owner audit

ElevenLabs uses AIML with possible fallback

AIML documents music_length_ms from 10,000 to 300,000 and warns actual length may differ. The audit reports music_length_ms = 60000 through AIML, not the direct ElevenLabs API; it may fall back to MiniMax Music 2.5. Fallback frequency and returned-model labeling are unverified. Picker estimates and public gateway prices are not actual customer charges.

Technical details

Choose by task, not by a universal ranking

These are editorial starting points, not measured comparisons. Check the exact model guide and current app access before using a model; linked catalogs may also contain historical entries.

Descriptive song or instrumental brief

Consider Suno Standard mode for a compact description, or inspect the Custom-mode contract when working with original lyrics. The gateway lists v5.5 and v5 separately.

Tradeoff: Suno's own voice/personalization features are not established here; exact version-specific maximum duration remains unverified.

Suno v5.5 guide

A lyric-led song

The gateway documents vocal songs and separate prompt/lyrics parameters. In the audited app, your main prompt is copied into both, so words you type may be treated as sung lyrics.

Tradeoff: There is no separate custom-lyrics input or independent style input in this app route. Keep descriptive instructions out of lyrics unless you intend them to be sung.

MiniMax Music 2.6 guide

Prompt-led composition through a gateway

AIML documents elevenlabs/eleven_music with a musical description and an optional length parameter. Its public API contract is distinct from the app's narrower audited request.

Tradeoff: The app may fall back to MiniMax Music 2.5. This route is not a guarantee that every returned result came from ElevenLabs.

ElevenLabs Music guide

Practical guidance

Create, refine and review

Replace every bracketed placeholder with authorized context. These are original templates, not generated-output demonstrations or generation-tested recipes. Copying does not submit anything or incur a generation charge.

Create a concise generation brief

Where to use it: AI Chat planning template; use only the resulting short brief in Suno Standard mode.

Write one original music-generation brief for [PURPOSE], using [MOOD], [INSTRUMENTS] and [ENERGY ARC]. Vocal preference: [VOCALS OR INSTRUMENTAL]. Return only a descriptive brief under 400 characters, with no artist imitation and no exact-duration promise. Do not add lyrics, API settings or explanations. Leave room for my edits. This is a Suno Standard-mode draft, not a MiniMax lyric input.

Check: Count the final brief after editing, then confirm the selected mode and any available instrumental control.

Refine a short original lyric draft

Where to use it: AI Chat planning template; review the resulting lyrics before using a lyric-led route.

Revise these original lyrics: [LYRICS]. Theme: [THEME]. Keep the meaning, use simple singable lines, and return only a short verse and chorus under 280 characters total, including section labels. Do not insert production instructions, artist names or claims about duration. The result is intended for a MiniMax main prompt that is also sent as lyrics, so every ordinary word may be sung.

Check: Read the returned words aloud and verify route-specific limits; the 280-character target leaves room under the documented Music 1.5 prompt ceiling too.

Review your listening notes

Where to use it: AI Chat review template; supply your own observations, not an assumption that chat can hear the track.

Compare [ORIGINAL BRIEF] with my listening notes: [AUDIBLE WORDS, INSTRUMENTS, STRUCTURE, UNWANTED SOUNDS AND MEASURED LENGTH]. Identify mismatches supported by these notes, not invented audio observations. Suggest one focused prompt revision and list changes better handled in an external editor. Do not claim to have heard the track, verified rights or identified the generating model. Flag missing evidence before recommending publication.

Check: Listen again to validate the suggested diagnosis. Do not rely on text analysis for an audio quality or rights determination.

Illustrative example

What a useful handoff could contain

An editorial example of structure, not a generated result, customer case study or measured outcome.

Illustrative instrumental brief

Purpose: background for a fictional workshop introduction. Direction: warm acoustic guitar, restrained percussion, a gentle rise and a resolved ending; avoid vocals. This is a creative intention, not a proven model result.

Generation record

Record selected model, mode and final input. Keep actual track length, audible words and returned-model identity as Unknown until evidence is available. A selected model is not proof that fallback did not occur.

Listening and edit handoff

Note where narration is masked, whether the ending is clean and which segment fits the intended edit. Timing adjustments, fades and mixing are external editing tasks unless separately verified in the app.

Technical details

Check the result before relying on it

Quality checklist

  • Listen to the full returned track for unintended words, artifacts and abrupt changes.
  • Measure actual length; never substitute the client's requested duration for the audio duration.
  • Compare the selected route, available result information and possible fallback without assuming provenance.
  • Review the applicable terms and permissions for your intended use; do not treat generation or a paid plan as automatic clearance.

When the result misses the brief

The track sings the instructions

On audited MiniMax routes, the main prompt is also lyrics. Rewrite the input as words you intend to be sung, or evaluate a route with the appropriate descriptive/instrumental controls.

The length does not match the brief

Check whether the route forwards any duration field. Plan an external edit if appropriate; repeatedly adding exact timing to a prompt does not create a missing app control.

The result does not fit under narration

Reduce the requested arrangement density for a new attempt or adjust the mix in an external editor. Verify by listening to the combined edit, not by asking chat to certify unheard audio.

Provider maximums, picker estimates and actual customer charges are different things. Read the pricing explanation.

Common questions

Before you start the workflow

Can I choose an exact 60-second track?

The audited interface has no duration selector. It requests 60 seconds, but some routes do not forward that value, and a forwarded length is not a guarantee of exact output. Measure the actual audio.

Does this include voice cloning or speech narration?

No voice cloning or speech-generation capability is established by this music workflow. Planning narration text in chat is different from generating speech. Check separately verified app features rather than inferring them from a provider's broader product range.

Can I release the result commercially?

This guide does not establish a license or clear rights. Review the applicable provider/gateway and EZ Ai Assist terms for the model, account and intended use, including supplied lyrics or other third-party material.

Evidence and review

Sources & boundaries

The linked technical references were reviewed on . Documentation and app availability can change.

App controls, forwarded parameters and fallback behavior come from the owner-supplied October 9, 2026 audit already documented in the website's music guides. Application code was not independently inspected. Official sources confirm their own contracts, not the app's successful execution. Prompts and examples are editorial; no audio was generated or listened to for this guide.

Opening EZ Ai Assist does not prefill these templates or automatically run the workflow. Choose the relevant mode and review your input before submitting it.

Continue exploring

View all use cases