From your first project to a finished dub
Make yourself at home.
Learn the editor at your own pace. Start with the practice project, or follow this guide with your own audio or video.
Updated 15 September 2026 · Beta user guide
1. Start here
You do not need to learn everything before trying your first dub.
Try the demo first — no credits needed
Sign in and open the demo project in your project list. Its guide walks you through listening, checking text, choosing voices, moving clips and exporting.
Use Show me where when you need help finding a control. Some steps depend on earlier actions, such as assigning voices.
Repeat a demo step or start again
Where available, choose Try this step again to reset that exercise. This is not available for every step, particularly when later work depends on it.
Restart demo resets the guide and edits in the practice project. Your own projects and credits are not affected.
Ready to use your own content?
- Create a project and upload audio or video you have permission to use.
- Review the detected speakers and original transcript before translating.
- Add a target language, check the translation and assign voices.
- Generate the dub. Listen, adjust individual lines and regenerate where needed.
- Export your finished audio, video or subtitles.
A desktop browser is recommended for detailed editing. Creating a real project requires a positive credit balance, even though transcription and translation do not themselves spend credits.
2. Upload and review
A good original transcript makes every later step easier.
Upload your file
Create a new project and choose your audio or video file. Common formats include MP4, MOV, WebM, MP3 and WAV. The standard duration limit is 30 minutes; account-specific limits may differ.
Dubedo detects the spoken language and speakers and separates speech from background audio. Overlapping voices, music and rapid cuts can make the result less accurate.
Check before translating
- Listen to the original audio and compare it with the transcript cards.
- Double-click a transcript card to correct its text. Press Enter to save or Escape to cancel.
- Check who is speaking, especially after scene changes. Click the person icon, speaker label and chevron at the top left of an original card to select another detected speaker. The pencil at the top right edits the text.
- When the original text is ready, move on to translation.
Editing text is not the same as generating audio. Do not assume that correcting the original transcript also updates translations or audio you already generated.
Changing a card’s speaker affects that segment and its translated cards, not the other original cards. Text, timing and explicit per-card voice overrides stay unchanged. If the effective voice changes, existing audio needs regeneration; changing the speaker itself does not spend credits. Extracted voice samples are not rebuilt. Speaker reassignment is unavailable in the prepared demo.
Upload and account limits
Plans have active-project and processing limits. Deleting a finished project can free an active-project slot, but does not reset daily usage limits.
Transcription, translation and voice previews may have daily limits even when they do not cost credits. Daily limits reset at midnight UTC. Follow the message shown in the app and see Pricing for current plans.
For longer uploads, contact hello@dubedo.com.
3. Translate and choose voices
Each target language has its own track. Each speaker can have a different voice.
Add a language
Choose Add language above the timeline. Review the translated cards for meaning, names and natural phrasing before generating audio. Use the language tabs to switch tracks.
A translation can take longer to say than the original. Shorten overly long lines without losing their meaning, then listen after generation. Matching text length does not guarantee matching duration.
Which target languages are available?
Dubedo currently offers these 30 target languages. Source-language detection is automatic; accuracy varies with language and recording quality.
- Arabic
- Burmese
- Chinese
- Danish
- Dutch
- English
- Finnish
- French
- German
- Greek
- Hebrew
- Hindi
- Indonesian
- Italian
- Japanese
- Khmer
- Korean
- Lao
- Malay
- Norwegian
- Polish
- Portuguese
- Russian
- Spanish
- Swahili
- Swedish
- Tagalog
- Thai
- Turkish
- Vietnamese
Choose a voice for each speaker
Open the speaker’s voice picker. Choose an available library voice or create one from the speaker’s extracted sample. Check the sample first: it should contain the intended speaker with as little other speech or background sound as possible.
Assign voices before generating the track. You can also override the voice for a single translated card without changing the whole speaker.
Clone from an uploaded recording or your microphone
- Open the voice library and choose the upload or recording option.
- Upload a reference, or allow microphone access and record clear speech in a quiet room.
- Play the reference recording to check it. This is your source sample, not the generated clone.
- Choose the clone settings and generate. Review the credit confirmation when shown.
- Listen to the generated voice preview below the reference section, then save the voice to your library.
Aim for 5–30 seconds of one clear voice. Uploaded references support up to 60 seconds and 10 MB. Use only voices you have the right and permission to clone.
Native-sounding or Closest to original?
Native-sounding aims for natural speech in the target language. Closest to original aims to retain more of the reference speaker’s delivery and accent.
Try the previews and choose by ear. Results vary with the recording, language and speaker; neither mode guarantees an exact match.
Create a voice from a prompt
In the voice library, choose the prompt option and describe the voice: for example, “a warm, calm adult voice with a British accent”. Generate, listen to the new preview and save a result you like. No reference recording is needed.
Initial creation costs credits. Eligible unchanged retries are shown in the form; changing the setup may start a new paid generation.
4. Generate and edit
Build the first version, then change only what needs attention.
Generate your language track
With your voices assigned, choose Generate and review the credit estimate. Wait for the job to finish before judging the result. Queueing and startup can add time before generation progress begins.
Listen to the dub with the background audio. Use individual card regeneration for local changes rather than starting the entire project again.
What do the card colours mean?
- Yellow / amber: pending changes. The card has changes that are not reflected in its generated audio. Regenerate it to hear the current version.
- Red: overlapping clips. Overlap is allowed. Listen for voices clashing and move or shorten a line if needed.
Move around the timeline
Click the time ruler to position the playhead. Drag an empty area left or right, or scroll over the timeline to move along it. Use the − / + zoom controls above the timeline, or Ctrl + scroll.
Drag a dubbed card horizontally to change when its audio starts. Moving it does not regenerate speech or spend credits. Playback can continue beyond the original recording when dubbed clips extend past it.
Edit text, delivery or the voice for one line
Double-click a translated card to edit its text. Save with Enter; use Shift + Enter for a new line, or Escape to cancel.
At the bottom of the card, choose + add instruction to describe delivery, such as “quiet and serious” or “excited”. Apply the instruction and regenerate to hear it. Instructions guide the result but are not exact controls.
The voice chip at the bottom lets you assign a different voice to that card. Text, voice and instruction changes require new audio and can incur a charge.
Add, split or delete a card
Add: after generating audio, use a small + between cards or after the last card. Enter a custom line, save it, check its voice and generate its audio.
Split: use the card’s split control to choose a break in the text. Review the resulting cards and timing before continuing.
Delete: use the trash icon, including on an empty custom card. You can also select a card and press Delete when you are not typing in a text field. Confirm the deletion.
Adding, splitting and deleting cards are not covered by Undo. Deleting a card does not refund previous generation costs.
Undo and redo: what is included?
Undo edit and Redo edit cover transcript text, segment speaker changes, translated text, clip timing and delivery instructions.
They do not restore audio takes or credits, change voice assignments, or undo adding, deleting or splitting cards. Generation and structural changes can clear the edit history. Leaving or reloading the project clears it too.
A dimmed Undo button means there is no available edit to undo, or an operation is still in progress. The information button beside it explains the scope.
Keyboard shortcuts
- Space
- Play or pause.
- Home
- Return to the beginning.
- Ctrl + scroll
- Zoom the timeline.
- Enter / Escape
- Save / cancel a text edit.
- Shift + Enter
- Insert a line break while editing.
- Delete
- Delete the selected card after confirmation, outside text editing.
- Ctrl/Cmd + Z
- Undo a supported timeline edit.
- Ctrl/Cmd + Shift + Z
- Redo a supported timeline edit.
When a text field or dialog has focus, shortcuts may apply there rather than to the timeline.
5. Listen and export
Check the mix, not just the words.
- Choose the target-language track you want to review.
- Use the timeline’s layer controls to listen to original speech, background audio and dubbed speech separately. For the dub, check that the dubbed layer is audible and the original speech is muted as intended.
- Listen through transitions, overlapping cards and the end of the last line. Regenerate pending changes.
- Choose Export, select the language tracks and format, and check the background-audio option.
Export formats
- Video: puts the dubbed mix into your video. Available for video projects.
- Audio: downloads an MP3, with or without background audio.
- Subtitles: downloads translated text and timing as SRT.
The export dialog lists eligible tracks. Review warnings about pending or missing audio before downloading. If you select multiple languages, your browser may ask permission for multiple downloads.
6. Credits and subscriptions
Check the estimate before generating. Prices and allowances are listed on the Pricing page.
What uses credits?
- Generating a language track: normally based on source duration, not only the spoken lines. Each generated target language is a separate operation.
- Creating a voice: initial creation from an extracted sample, upload, microphone or prompt costs credits.
- Regenerating: unchanged results have a limited number of free retries. Changes to text, voice or delivery instructions are not an unchanged retry.
- No generation charge: transcription, translation, moving cards and replaying existing audio. Daily usage limits can still apply to processing operations.
See current prices and allowances. The confirmation or generation form shows the applicable cost and available free attempts.
Free retries and new previews
Generated clips currently include three free retries with the same text, voice and instruction. Voice-creation forms also show eligible free retries for an unchanged setup.
Playing an existing preview does not generate anything. Requesting a new preview is a new processing operation; check the form’s cost and retry indicator. Do not assume that changing settings or making a new voice preserves the old free attempts.
Running out of credits
You can still open and edit existing projects with a zero balance. New real projects require a positive balance, and paid generation needs enough credits for the operation.
Where offered, partial generation lets you generate some of a track and continue after adding credits. Review which lines are actually ready before exporting.
Open Pricing to top up or change your plan, complete checkout, then return to the editor. If the balance still looks stale, reload once and contact support if the payment has completed but credits remain missing. Do not repeatedly pay to resolve a display issue.
Rollover, top-ups and cancellation
Unused subscription credits roll over up to a total subscription balance of twice your monthly allowance after renewal. Purchased top-ups are separate from that cap and do not expire while your account remains open.
Manage your subscription through the billing portal. Cancellation takes effect at the end of the paid period. Read Terms for cancellation conditions and Pricing for plan details.
7. Troubleshooting
Start with the symptom you can see or hear.
The transcript or speakers are wrong
Compare the original recording with the cards at the point the result changes. Background music, overlapping speech and scene cuts can confuse speaker detection. Correct text before translating; renaming a speaker alone does not repair incorrectly grouped speech.
Use the speaker selector on an original card to correct a misassigned segment. If a single card contains multiple people talking, this selector cannot split their speech: send support the project link and timestamps if you cannot correct it. A clearer source or shorter excerpt can help on a new attempt.
The dub is silent, or I hear both languages
Select the intended language track and check the mute controls for original speech and dubbed audio. Confirm that audio generation has completed and that the playhead is over a generated clip.
If the controls look correct, pause and play again. Once saving and generation have finished, try reopening the project. Report recurring playback problems with the browser and affected timestamp.
A line is too long or overlaps another
A red border marks overlap, not a failed generation. Listen first. Move the card if there is room, or shorten the translation and regenerate while preserving its meaning. A delivery instruction can help, but cannot guarantee an exact duration.
Voice recording or its preview does not work
Allow microphone access in your browser and select the intended input device. Play the recorded reference before generating the clone.
The generated clone has a separate preview below the reference section. If an upload is rejected, check the format, size and duration; try exporting a valid WAV or MP3 instead of just renaming the extension.
Processing is queued or seems stuck
A job can wait for capacity or startup before processing begins. Follow any retry guidance shown in the app and avoid submitting the same job repeatedly.
If progress remains unchanged unexpectedly or an error appears, note the message and approximate time and send the project link to support. Do not delete the project before reporting it; it can help us investigate.
Export is missing a track or warns about changes
Check that the intended track has finished generating. Pending text, voice or instruction changes need regeneration to be reflected in audio. With partial generation, some lines may still have no audio.
Audio-only projects do not offer video export. If a download is blocked, check your browser’s download permissions.
8. Help and your data
Tell us where something went wrong so we can investigate.
Contact support
Email hello@dubedo.com with the project link, what you expected, what happened, the approximate time and your browser. For audio issues, include the timestamp. A screenshot of the error is helpful.
Do not send passwords, API keys or payment-card details.
Delete a project or your account
Delete individual projects from the project list. Account deletion is available in the profile panel and is permanent; it removes your projects and voice library. Recreating the account does not restart the free trial.
For retention periods, backup handling and records we may need to keep, read the Privacy policy. Use project deletion, not account deletion, if you just want to start a new edit.