Upload and Extend Audio continuation API

Provide an accessible audio source and choose the point where the next musical section should begin. Upload and Extend Audio uses upload_url with a source-based or custom-parameter workflow. Custom continuation prepares mv, instrumental, a positive continue_at value, style and title, adding prompt for vocal creation. Listen to the completed result with the original and review how the new section develops the music.

Listen to the returned continuation around the selected starting point and through the new passage, then select the candidate that fits the source and the intended development. Save its audio URL, returned identifiers and actual duration together with the source recording. This gives you the recording and provenance needed to arrange the passage in an editing project or prepare a supported follow-up music request.

Submit model: upload-and-extend-audio with input.upload_url and default_param_flag. For custom parameters, include instrumental, positive continue_at, mv, style and title, plus prompt for vocals. Retain task_id and query music details for the completed continuation candidates.

Use default_param_flag: false when you want to start from the uploaded source with the source-based workflow. Choose true when you want to set the next passage yourself, and supply instrumental, a positive continue_at, mv, style and title. Include prompt when instrumental is false. Set continue_at within the uploaded recording’s duration and describe the passage that should follow that point.

Use an accessible upload_url; the online uploader has a 50 MiB file limit. The task documentation gives an eight-minute duration limit, or one minute for V4.5 All. With default_param_flag: true, provide instrumental, positive continue_at, mv, style and title; add prompt when instrumental is false. With false, upload_url is the required source and other accepted fields refine the request. The task documentation lists V4 prompt/style limits of 3,000/200 characters and V4.5, V4.5 All, V4.5 Plus and V5 limits of 5,000/1,000. Titles use 80 characters for V4 and V4.5 All, and 100 for V4.5, V4.5 Plus and V5.

Choose the source recording and retain a directly downloadable audio URL. Follow the documented duration and online file-size limits. Prepare the point to continue and, for custom work, the next lyrics or music direction, style and title.

Read this page’s dynamic pricing section for the current task rate per processing request. Source music generation and other follow-up operations use their own rates. Review the current price before submitting, and retain the completed result and task association when planning additional work.

This task starts from an uploaded audio URL. Extend Music instead starts from an existing generated audio_id, while Upload and Cover Audio prepares a restyled interpretation. Within this task, default_param_flag selects source-based or custom parameters, and instrumental determines whether the custom prompt supplies vocal text.

Listen around the selected continuation point and through the new section. Compare the melody, rhythm and applicable lyric delivery with the source and the custom brief. Use negative_tags to describe musical elements to reduce and vocal_gender m or f as a vocal preference when applicable. style_weight guides style adherence, weirdness_constraint guides creative deviation and audio_weight sets the documented audio-feature balance; their documented range is 0–1 with up to two decimals. Apply an existing persona_id where the selected workflow supports it. Save the selected result and original source together when preparing further work.