Audio Generation

Live catalog

Supported audio models

Choose a model before you generate. Availability and modes are loaded from the current AnyCap catalog.

View all modelsInstall the CLI

Loading live models…

Audio generation

Audio is a generation capability distinct from music. The selected model may expose one or more of these mode values; inspect the live catalog before choosing one:

Mode	Use it for
text-to-audio	Create an audio scene, speech, or dialogue from a prompt
audio-to-audio	Guide a new result with supported audio references
image-to-audio	Guide an audio scene with a supported image reference

# Print the modes advertised by every audio model right now.
anycap audio models \
  | jq -r '.models[] | .model as $model | .operations[]?.modes[]?.mode | [$model, .] | @tsv'

anycap audio models
anycap audio models <model-id> schema --mode text-to-audio
anycap audio generate \
  --prompt 'A calm narrator says: "Welcome to AnyCap." Quiet studio ambience.' \
  --model <model-id> \
  --mode text-to-audio \
  -o welcome.mp3

These are generation workflows, not claims of a general-purpose audio editor. Confirm formats, sample rates, reference limits, subtitles, and voice controls against the selected live schema.

Output metadata

Audio outputs can include local_path, mime_type, duration, byte size, subtitle data, usage, and diagnostic IDs. If one local download fails after provider generation succeeds, that output reports its own error instead of hiding the successful provider result.

For composed tracks, see Music generation.