Smart Uploads: Mastering Pre-processing Settings for MP4 and MOV
Getting your video into Stra.ai is simple — drag, drop, configure, and hit Create. But the settings you choose in that popup before you click the button have a bigger impact on your results than most people realize. This guide walks through every option so your first project comes out right.
File requirements before you upload
Stra.ai accepts MP4 and MOV files. Before you upload, check these two limits:
Minimum file size: 5MB. Files smaller than this will return an error. If your clip is very short, it likely falls under this threshold — try exporting at a higher bitrate or combining clips before uploading.
Maximum file size: 1GB. Files over 1GB will not process. If your video is longer than roughly 30–40 minutes at standard quality, consider splitting it into parts and processing them separately.
Outside of those two limits, Stra.ai handles the rest. No need to transcode, reformat, or adjust frame rates before uploading.
A credit-saving tip before you upload
This matters especially for longer videos: Stra.ai charges credits for every minute of video processed ,including silent sections, B-roll, and any part of the video with no dialogue.
If your video is 10 minutes long but only 3 minutes of it contains actual speech, you are paying to process 7 minutes of silence.
Before uploading, trim your video down to the dialogue sections only using any editor (CapCut, DaVinci Resolve, Premiere, whatever you have). Export the trimmed version and upload that instead. You will use fewer credits, get faster processing, and have a cleaner project to edit.
Note: If you are using a YouTube link this is not possible , you get the full video as-is. More on that below.
Uploading from your computer
Inside the Create popup, drag your file onto the upload area or click to browse. Once the file loads the popup will reveal the full settings panel.
If you are creating an AI Subtitles project, the settings are straightforward: source language, target language, optional SRT upload, and translation prompt. Those are covered in the Dashboard guide.
If you are creating an AI Dubbing project, there are several additional settings worth understanding. Read on.
Uploading from YouTube (AI Dubbing only)
In the AI Dubbing popup, you will see a YouTube URL input field alongside the file upload area. Paste your link and Stra.ai will download the video automatically.
YouTube downloads take longer than direct file uploads ( a 10-minute video can take up to a minute before the credit estimate appears. This is normal. Wait for the credit calculation to complete before clicking Create.
One important thing: make sure your speaker detection is set to Auto Detect before clicking Create. This is the most reliable configuration for YouTube uploads and avoids a known issue with manual speaker counts. More on that below.
Source language — get this right
Select the language being spoken in your original video. If you get this wrong — for example setting Spanish when the video is in Korean ) the transcription will produce gibberish and the entire project will fail.
If you are not certain of the language, use Auto Detect. It is accurate for most common languages.
Generate translation first (AI Dubbing only)
This toggle outputs a subtitle file from your video without generating any audio. Use it when you want to review and edit the translated script before committing to full voice generation.
The credit cost is the same whether this toggle is on or off , you are not saving credits by using it. What you are saving is the time and effort of re-generating audio after discovering a translation error. For longer videos or content where accuracy matters, turning this on first is a good habit.
Background audio separation (AI Dubbing only)
This setting tells Stra.ai how to handle the audio in your video before dubbing. You have four options:
Roformer — General purpose model. Good default for most videos with a mix of voice and background audio.
MDX-Net — Music removal. Best for videos where there is a music track underneath the dialogue. Separates voice from music more cleanly.
Clear voice. Best for interview-style or talking-head videos where the main audio is just speech with minimal background noise.
No separation — Voice only. Use this if your video has no background audio at all ,. a clean voiceover recording, a podcast-style video, or a screen recording with narration. Skips the separation step entirely for faster processing.
Choose the option that most closely matches your source audio. When in doubt, Roformer is the safest general pick.
Number of speakers (AI Dubbing only)
This setting tells the AI how many different voices are in your video so it can separate and assign them correctly.
Always use Auto Detect. It is the most reliable option and works well for most content. Manually selecting a speaker count triggers a known issue where the AI assigns dialogue to the wrong speakers — and that is significantly harder to fix in the editor than just letting Auto Detect handle it from the start.
Leave the dropdown on Auto Detect and move on
Voice model (AI Dubbing only)
Stra.ai offers two voice generation engines:
ElevenLabs — Natural-sounding voices with voice cloning. Best for content where the dubbed voice should closely match the original speaker's tone and character. Good for vlogs, interviews, and personal content.
Gemini TTS — Best for emotion and tone control. Ideal for scripted content, narration, and anything where you want precise directorial control over how the AI delivers each line. Note that the directing prompt feature only works when Gemini TTS is selected.
Translation prompt
Choose a style preset to guide the tone of the translation — or write your own custom prompt if none of the presets fit your content.
Available presets: Live Sports Commentary, Casual Drama/Show Style, Formal News Report, Teen/Gen Z Style, Child Friendly Style.
If you are unsure, leave it blank. Stra.ai will produce a neutral, accurate translation without a prompt.
What happens after you click Create
Once you hit Create AI Dubbing, Stra.ai begins processing in stages. Your project will appear in the dashboard and update its status in real time:
Separating voice from background
Transcribing
Translating
Preview ready
You do not need to stay on the page. You can come back when your project is ready to edit. Go do something else and come back later if you need to but it does not take too long.
If something goes wrong
If the transcription fails, no confirmation email will arrive. Go back to the dashboard and check your project status . If it failed, the project will show an error and the editor will open with empty segments and only the separated audio.
The good news: if a project fails due to a processing error, Stra.ai automatically refunds the credits. You do not lose anything. Report the issue directly to the Stra.ai support team and they will investigate.
If you ran into an upload error rather than a processing error, check the troubleshooting guide for the most common causes and fixes.
What to do next
To learn how to edit your subtitle project → AI Subtitles Overview
To learn how to edit your dubbing project → AI Dubbing Overview
To get better translation results → The Power of Custom Prompts
To understand credit costs → Token Economics Guide
→ Continue here: The Power of Custom Prompts — How to Get Netflix-Standard Subtitles
