This is the largest group of editing jobs Versely runs, and most of the trouble in it comes from conflating three different operations. Replacing audio is mechanical: a new track goes over the old one. Dubbing is translation plus performance. Lipsync is a video job that happens to be triggered by audio — it repaints a mouth to match a waveform.
Which of the three you need depends on one question: is there a face on screen? If there is not, everything here is an audio swap. If there is, the audio decision constrains the video, and doing them in the wrong order means generating the same clip twice.
The order to run these in
These jobs are cheap in the right sequence and expensive in the wrong one. This is the order that avoids re-rendering work you have already paid for.
1. Get the words down first
A transcript or a script is the input to everything else in this cluster — dubbing, voiceover, captions and lipsync all read from it.
How to do it →2. Choose the voice before the picture
Stock voice, cloned voice or the original performer. This decision constrains everything downstream, so make it before you generate video.
How to do it →3. Sync the mouth only if a face is on screen
Lipsync is the expensive step and the only one that re-renders picture. Skip it entirely for voiceover-over-b-roll.
How to do it →4. Put music underneath last
Score against the finished voice track so the mix ducks in the right places, rather than fighting a bed that was chosen first.
How to do it →
Every job in this cluster
Each one names the real Versely agent tool that performs it, so you can see which jobs share a mechanism and which are genuinely different work.
Add Music to a Video
A soundtrack under your voice, not over it.
attach_audio_to_videoAdd a Voiceover to a Video
Type the script. Get a narrated video back.
generate_speechReplace the Audio in a Video
Mute the original. Put your own sound in its place.
attach_audio_to_videoDub a Video Into Another Language
Same voice, same face, new language.
dub_videoTranslate a Video Into Another Language
Just the words, translated — lighter than a full dub.
dub_videoLipsync a Photo or Video to Audio
One photo, one audio file, one talking clip.
generate_lipsyncAdd Multi-Speaker Dialogue to a Video
A whole cast, one API call.
generate_multi_speaker_speechClone Your Voice for Videos
Record once. Reuse the voice forever.
clone_voice_from_audioChange the Voice in a Video
Same performance, different voice.
change_voiceExtend a Music Track
Not long enough? Keep it going.
extend_musicIsolate Vocals From a Track
Strip the instrumental. Keep the voice.
isolate_audioGet a Video Transcript
Turn spoken words into a written script.
transcribe_audioAdd a Cover Song to Your Video
Same song, reimagined arrangement.
cover_musicSplit a Song Into Stems
One track in. Two usable stems out.
separate_music_vocalsWrite Custom Lyrics for Background Music
Get the words right before you commit to a full track.
generate_lyricsCreate a Music Video From a Song
Turn a finished track into something postable.
create_music_videoThe agent tools behind them
18 tools from Versely's real agent surface do all 16 of these jobs. Ask for the outcome in plain English and the agent picks the tool.
attach_audio_to_video
Attach an audio file to an existing video. Use when the user has a generated/uploaded video AND a separate audio track (music, voiceover, SFX, generated speech) and wants them combined. Two modes: (A) 'replace' (default) — strips the video's original audio and substitutes the supplied audio. Pick this when the user says 'replace the sound', 'mute the original', 'put this music on the video', or any time the supplied audio should be the only audible track. (B) 'mix' — keeps the original video audio AND mixes the new audio on top, with separate volume controls. Pick this when the user wants background music UNDER existing voice/SFX (e.g., 'add background music', 'put music behind my talking video'). Video stream is copied losslessly (no re-encode of visuals). Fast. Costs a flat credit fee regardless of length.
generate_music
Generate music or song audio from a text prompt. Use this when the user asks for music, songs, background scores or melodies.
generate_speech
Generate speech audio from text (Text-to-Speech). Use when the user needs a voiceover, narration, or spoken audio for their content. Perfect for UGC workflows, ad scripts, and podcast clips. Use `emotion` and/or `style_instructions` to make the delivery expressive and match the content's mood instead of defaulting to a flat read — pick whichever the target model supports.
dub_video
Dub a video or audio file into another language — full AI dubbing (voice cloning + optional lip-sync), not simple TTS narration. Source must already be Versely-hosted media (a URL from a prior generation, upload, or output — e.g. *.versely.studio); external links (YouTube etc.) aren't supported yet. Runs asynchronously: submits the job and returns a project_id — poll get_dub_status or check list_dubs for completion, don't wait in this turn. Two engines: 'elevenlabs' (default, audio-cloning dub, works on audio or video, supports trimming via start_time/end_time, up to 30 min) and 'heygen' (video-only, lip-synced translation, up to 8 min, no trimming, and only supports a subset of languages).
get_dub_status
Check the status of a dubbing project by id — poll this after dub_video while the job is in flight (status: queued/dubbing/completed/failed). Returns the output URL(s) once complete.
list_dubs
List the user's dubbing projects, newest first. Use when the user asks to see their dubs or find a project they don't have the id for.
generate_lipsync
Generate a lipsync video — animate a face image to speak with provided audio. Use when the user wants to make a person in an image speak, create a talking avatar, or lip-sync audio to a face.
generate_multi_speaker_speech
Generate multi-voice dialogue/podcast audio in one call (Gemini multi-speaker TTS) — each speaker alias gets its own distinct voice, spoken in the order their lines appear in `text`. Use for scripted conversations, interviews, or podcast-style clips with 2+ distinct voices instead of chaining multiple generate_speech calls.
clone_voice_from_audio
Clone a reusable Cartesia voice from an audio sample. Returns a voice_id usable with generate_speech (Cartesia models) and change_voice. Use when the user wants to create a voice from a recording rather than pick from the existing catalog.
change_voice
Convert an existing audio clip's voice to a different Cartesia voice, keeping the delivery/timing (Cartesia voice changer). Use when the user has an audio recording and wants it re-voiced rather than re-typed and re-spoken.
list_cloned_voices
List the user's cloned voices across both Cartesia and Inworld, newest first (capped at 30). Use to find a voice_id for generate_speech or change_voice, or to show the user what they've cloned.
transcribe_audio
Transcribe an audio clip to text (Cartesia ink-whisper speech-to-text). Use when the user wants a spoken clip turned into a written transcript.
extend_music
Extend an existing Suno-generated track with more music, continuing from (or near) a point in the source track. Use when the user wants a previously generated track made longer. Requires the source track's audioId (get it from get_generation_history / fetch_user_generations for a prior generate_music result).
isolate_audio
Isolate/extract vocals from an audio clip (vocal isolation), removing background music/instrumentation. Use when the user wants a clean vocal track from a mixed recording.
cover_music
Generate a Suno cover/reinterpretation of an existing audio track from its URL (new style/arrangement over the same song). Use when the user wants a cover version of a track they have a URL for (their own upload or a prior generation), not a from-scratch new song.
separate_music_vocals
Split a previously generated Suno track into vocal/instrumental stems. Requires the source track's taskId and audioId (from a prior generate_music / extend_music result).
generate_lyrics
Generate standalone song lyrics from a text prompt/theme (Suno lyrics generation) — does not produce audio, just lyric text. Use before generate_music when the user wants to review/edit lyrics first, or when they just want lyrics.
create_music_video
Render a simple music video (waveform/visualizer over the track) for a previously generated Suno track. Requires the source track's taskId and audioId (from a prior generate_music result).
Models this cluster runs
Jobs in this cluster that re-render pixels name a specific model. Prices are in Versely credits.
What these jobs cost
Every job here bills in credits. These are the scenarios whose formulas govern the arithmetic.
What does an AI video with sound cost?
flips the per-second rate with an audio toggle rather than adding a separate fee
Per 1,000 characters, rounded upWhat does an AI voiceover cost for my script?
meters the characters you type, not the seconds you get back
Flat per job, language-invariantWhat does dubbing one video into 10 languages cost?
bills one flat charge per job and ignores the target language entirely
Linear per secondWhat does a 30-second AI talking head cost?
bills a flat rate for every second of output, so length is the only lever
Additive across four different metersWhat does a finished 30-second AI ad cost end to end?
sums four unrelated meters, so no single rate predicts the total
Questions about this cluster
What's the difference between dubbing and changing the voice?+
Dubbing translates the words into another language and re-performs them. Voice change keeps the words exactly as they are and swaps who appears to be saying them. Reaching for the wrong one produces either an untranslated clip in a new voice, or a translation nobody asked for.
Do I always need lipsync after replacing audio?+
Only when a talking face is visible. Narration over b-roll, product footage or slideshow content needs no lipsync at all — and skipping it saves the most expensive job in this cluster.
Does generated music need a licence?+
Music generated inside Versely on a paid plan comes with commercial rights, which is the point of generating it rather than sourcing it. Music you upload yourself carries whatever licence you obtained for it, and Versely cannot grant rights it does not hold.
Make the material first
AI Music Generator
Describe a vibe. Get a song. Keep the rights.
AI Video Generator
Text-to-video, image-to-video, and story-to-video in one place.
AI Voice Cloning & Text to Speech
Your voice. Any language. Any script.
AI Lipsync Generator
Text, audio or video in. Pixel-perfect talking head out.
AI UGC Video Generator
UGC ads at the speed and price of a prompt.
The rest of the pipeline
Captions & On-Screen Text
The last thing you add and the first thing anyone reads.
Aspect Ratio & Resize
Six jobs, one decision: what leaves the frame.
Trim & Cleanup
Do these before you pay to render anything.
Upscale & Extend
The only jobs that add something the file never had.
Merge & Assemble
Where the pieces stop being clips and start being a video.
Ask the agent for it in plain English
The all-in-one AI studio for creators. 60+ models for video, image, voice, music and lipsync in a single app.