Photo + Audio starts with a recording you already have. The useful first question is not “Which voice should I choose?” but “Which part of my existing recording do I want to build a visual performance around?” Keep that distinction clear before you select a file.
Bring a recording, not a promise of one
The workflow pairs your portrait with supplied audio. It is not a lyrics-to-song service or a voice-cloning workflow. Writing lyrics, recording them and preparing a visual performance are separate steps, even when they belong to the same creative project.
Use a recording you own or have permission to process. If another person performs on it, consider their voice and performance permission separately from rights to the instrumental or composition.
Choose a phrase that stands on its own
Listen for one complete thought or rhythmic section. Starting with a fragment can make it harder to judge the intended delivery. Keep the section short enough that you can listen critically more than once rather than repeatedly moving the bounds without a goal.
For a conventional 4/4 passage, one bar lasts 240 divided by the BPM in seconds. Eight bars at 120 BPM would therefore take 16 seconds. That is a planning calculation, not automatic beat detection: your recording may use different meter, pickups, pauses or tempo changes. The studio follows the actual media duration.
Check format, size and playback separately
The audio preview accepts MP3, WAV, M4A and OGG, up to 30 MB. The selected interval must fit inside the recording and be no longer than 30 seconds. A large WAV file does not become smaller just because you select a shorter interval.
If your browser does not play a file, try a properly converted local copy in a supported format. Renaming an extension is not conversion. Listen to the converted copy all the way through the section you plan to use; do not assume the export preserved its timing.
Do not mistake loudness for clarity
Audition at a comfortable volume and listen for clipped peaks, competing voices or sections that begin in the middle of a word. The current studio previews the recording you supplied; it does not repair, master or generate a replacement for it.
If you need to make edits, do them deliberately in your own audio editor and keep the source. Leave enough context to hear the phrase and avoid making a change simply because a waveform looks more impressive.
Know what happens in this version
Selected audio is previewed through a temporary browser-local URL. Within the current page session, switching to the reference-video tab preserves the audio selection and its clip boundaries. Leaving or reloading the page clears the local selection; it is not an account upload.
Provider-dependent rap generation is not connected. No generated voice, finished video, task progress or credit charge is simulated. You can prepare the inputs now and return to verified release notes when an integration becomes available.