Sonilo

Analyze a publicly reachable video URL and return a creative brief: a time-aligned segment plan and one or more prompts, derived from the footage itself. mode picks which brief — "both" (the default) writes a music-direction brief in segments/variations AND a sound-design brief in sfx_segments/sfx_prompt in one call; "music" writes only the scoring plan for text_to_music, video_to_music and video_to_video_music; "sfx" writes only the sound design for text_to_sfx, video_to_sfx and video_to_video_sfx. All three cost the same. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests it. Video duration must not exceed 480 seconds. prompt is optional guidance for the analysis; in both mode it steers the music brief, and the sound design follows the footage. variants_num (1-5, default 1) is how many independent briefs to author for the same video, billed per brief; the sound-design prompt is authored once regardless. Use get_generation_task to retrieve the result.

analyze_video

Dub a publicly reachable HTTPS video URL into one or more target languages, returning one dubbed video per language. This starts a paid Sonilo generation, billed per language, and may incur charges, so call it only after the user explicitly requests generation. video_url must use https specifically — the dubbing pipeline requires it. languages is a string holding a JSON array of language codes, e.g. '["ja", "es"]', from: en, zh_cn, ja, ko, pt, pt_br, es, es_419, de, fr, it, ru, th, ar, tr, vi, id, ta, ml, kn, gu, pa_in, sd_in, hi; omit it to get the default ["zh_cn", "es", "fr"]. pt_br is Brazilian Portuguese and es_419 is Latin American Spanish; plain pt and es give unqualified Portuguese and Spanish, so pick the regional code when the region matters. ar is likewise unqualified Arabic, not a country dialect. pa_in and sd_in are the Indian variants of Punjabi and Sindhi, the only ones available; there is no Pakistani variant of either. Set ducking=true to duck the background music/effects bed under the dubbed voice while it speaks; off by default (the bed is always kept, at a constant level), and free either way. lipsync defaults to true: the speaker's mouth is re-rendered to match the dubbed speech. Set lipsync=false to leave the picture completely untouched instead — the video comes back at its original resolution and frame rate rather than being re-rendered, and only the audio is replaced, so the mouths keep moving to the original language. Use it for footage with no on-camera speaker, or when preserving the exact original picture matters more than matching lip movement. subtitles is an optional JSON object mapping each requested language code to a public https URL of a .srt or .vtt script in that TARGET language, e.g. {"ja": "https://…/ep1.ja.srt"}; the dub speaks those lines verbatim instead of translating the source, and every requested language needs exactly one script. Every script is validated before anything is billed, including that its text matches the declared language. export_srt (default false, needs subtitles) additionally returns each language's lines re-timed against its dubbed audio as subtitles[language] on the task. Video duration must not exceed 300 seconds. Free preview: a self-serve account's first single-language video translation is free — the video is trimmed to its first 15 seconds and dubbed at no charge, and the task's trial_preview quotes what the full video costs. Several languages or a script are billed as usual. Use get_generation_task to retrieve the result — a dubbed video URL per requested language.

dubbing

Lower music automatically while speech is present. Both inputs must be publicly reachable HTTPS audio or video URLs. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests processing. Use get_generation_task to retrieve the result.

audio_ducking

Generate a complete soundtrack — music AND sound effects, mixed together — for a publicly reachable HTTPS video URL, returned as audio plus separate music/sfx stems. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests generation. Set preserve_speech to keep the video's own speech in the mix (no extra charge). ducking (default off, free) keeps the video's own voice and dips the generated music under it; pass ducking=true to opt in. output_format defaults to wav; set it to m4a for AAC or mp3 for 320 kbps MP3 if the user specifically asks for one of those. segments (optional) is a JSON-encoded array string of up to 30 {"start": float, "end": float, "prompt": str} objects giving per-segment SFX descriptions for the sound-effects layer: the first start must be 0, segments must be contiguous (each end equals the next start), every end must exceed its start, every prompt must be 1-200 chars, and the last end must not exceed the video duration. Omit it to let Sonilo choose sound effects for the whole video. Use get_generation_task to retrieve the result.

video_to_sound

Generate a complete soundtrack — music AND sound effects, mixed together — for a publicly reachable HTTPS video URL and return a NEW video with that soundtrack mixed in, plus separate music/sfx stems. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests generation. By default the returned video's audio is the generated music and sound effects ALONE -- the source video's own audio is removed. Pass keep_original_sound=true to keep the source audio, statically mixed with the generated music (free), adding ducking=true to dip the music under the source voice instead, or preserve_speech=true to keep only the isolated speech (also free). segments (optional) is a JSON-encoded array string of up to 30 {"start": float, "end": float, "prompt": str} objects giving per-segment SFX descriptions for the sound-effects layer: the first start must be 0, segments must be contiguous (each end equals the next start), every end must exceed its start, every prompt must be 1-200 chars, and the last end must not exceed the video duration. Omit it to let Sonilo choose sound effects for the whole video. Use get_generation_task to retrieve the result video URL.

video_to_video_sound

Score a publicly reachable HTTPS video URL and return a NEW video with generated music mixed in. By default the returned video's audio is the generated music ALONE -- the source video's own audio is removed. Pass keep_original_sound=true to keep the source audio, statically mixed with the generated music, adding ducking=true to dip the music under the source voice instead, or preserve_speech=true to keep only the isolated speech. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests generation. Video duration must not exceed 360 seconds. Accepted formats: mp4, mov, m4v, or webm carrying H.264, H.265/HEVC, VP9, or AV1 video — the source picture is copied into the result without re-encoding, so animated gif and VP8 webm are not supported. segments (optional) is a JSON-encoded array string of {"start": float, "prompt": str} objects describing how the music should develop over time; the first start must be 0 and the last segment must leave room before the video ends. Omit it to let Sonilo score the whole video. prompt_influence (0-1, default 0.5) sets how strongly generation follows the prompt; omit it unless the user asks for stricter or looser prompt adherence. Use get_generation_task to retrieve the result video URL.

video_to_video_music

Add sound effects to a publicly reachable HTTPS video URL and return a NEW video with the SFX mixed in. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests generation. segments (optional) is a JSON-encoded array string of up to 30 {"start": float, "end": float, "prompt": str} objects giving per-segment SFX descriptions: the first start must be 0, segments must be contiguous (each end equals the next start), every end must exceed its start, every prompt must be 1-200 chars, and the last end must not exceed the video duration — e.g. '[{"start": 0, "end": 2.5, "prompt": "footsteps on gravel"}, {"start": 2.5, "end": 5, "prompt": "door slam"}]'. Omit it to let Sonilo choose sound effects for the whole video. Use get_generation_task to retrieve the result video URL.

video_to_video_sfx

Generate licensed music from a text prompt. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests generation. duration is optional: leave it out and a length is inferred from the prompt (90 seconds if none can be). output_format defaults to m4a; set it to wav for 16-bit PCM or mp3 for 320 kbps MP3 if the user specifically asks for one of those. Pass stems=true to also get the track split into four separate instrument tracks — drums, bass, vocals and other — at no extra charge; it adds a few minutes to the wait, so set it only when the user asks for stems or separated tracks. Use get_generation_task to retrieve the result when processing finishes. If a stems result comes back with stems_error instead of (or alongside a partial) stems, only the separation failed or was skipped: the music itself generated fine, the audio URLs are valid, and the user was not charged for the separation — report it as a missing extra, not as a failed generation.

text_to_music

Generate licensed music that matches a publicly reachable HTTPS video URL. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests generation. Set preserve_speech to also separate the video's own speech/vocals and mix them against the generated music (no extra charge). Pass ducking=true to also get a free extra track with the generated music automatically lowered while the video's own voice is present — it is OFF by default. output_format defaults to m4a; set it to wav for 16-bit PCM or mp3 for 320 kbps MP3 if the user specifically asks for one of those. prompt_influence (0-1, default 0.5) sets how strongly generation follows the prompt; omit it unless the user asks for stricter or looser prompt adherence. Pass stems=true to also get the generated music split into four separate instrument tracks — drums, bass, vocals and other — at no extra charge; it adds a few minutes to the wait, so set it only when the user asks for stems or separated tracks. It splits the generated music, never the video's own audio. Use get_generation_task to retrieve the result — the audio URL, plus vocals/mux URLs when preserve_speech was set, ducked URLs when ducking ran, and stems URLs when stems was set. If a stems result comes back with stems_error instead of (or alongside a partial) stems, only the separation failed or was skipped: the music itself generated fine, the audio URLs are valid, and the user was not charged for the separation — report it as a missing extra, not as a failed generation.

video_to_music

Generate a sound effect from text. This starts a paid Sonilo generation and may incur charges, so call it only after the user explicitly requests generation. duration defaults to 8 seconds. audio_format defaults to aac; set it to wav, mp3, or flac only if the user specifically asks for that format. Use get_generation_task to retrieve the result when processing finishes.

text_to_sfx