Upload the MP4 as it is
An MP4 file is a container. It usually holds H.264 or H.265 video together with an AAC audio track. To make subtitles, only the audio matters, and RocketScribe pulls it out of the video on its side. You upload the same MP4 you plan to publish, and nothing needs converting beforehand.
Because the subtitles come from the same file as the video, the timing lines up from the first frame. That is the main advantage over sending a separately exported audio file.
Phone videos, screen recordings, camera footage and finished exports from an editor all work. Uploads are sent in parts, so if the connection drops on a long video, Retry continues where it stopped instead of starting over.
MP4 to SRT for YouTube
YouTube creates automatic captions for many videos, but you cannot correct them before they go live, and they can lag on names and technical terms. Uploading your own SRT gives you captions you have already checked.
In YouTube Studio, open the video's Subtitles page, add or choose the language, and upload the file with timing. YouTube accepts SRT and VTT. Captions also make a video usable with the sound off, and the text gives viewers who rely on captions the full content.
If you post the same video on other platforms, check whether they accept an SRT upload. Many do, and the same file works in each place.
Importing the SRT into Premiere Pro, DaVinci Resolve or CapCut
Video editors can import an SRT file as a caption or subtitle track. Premiere Pro, DaVinci Resolve and the CapCut desktop app all support this. Once the track is on your timeline, the editor lets you change fonts, colours and position, and you can burn the captions into the picture when you export the video.
This is the right place for styling. RocketScribe produces the text and the timing only. It does not burn subtitles into the video and has no subtitle styling options.
Import the SRT into the same project that holds the video you uploaded. If you have since trimmed the start of the edit, move the subtitle track to match, since every cue is timed from the start of the original file.
How cues are timed and what to check
RocketScribe splits the transcript into short cues of up to two lines. A new cue starts at the end of a sentence, at a longer pause or when the speaker changes. These rules are fixed. You cannot set a maximum number of characters per line or a cue duration.
Before you export, play the video's transcript in the synced player and fix any line that reads wrong. The SRT is built from the corrected text and keeps the original timing.
- Names, brands and jargon are the usual errors, since there is no custom vocabulary.
- Music played at the same level as speech lowers accuracy. Videos with a clean voice track work best.
- For videos with two or more people, speaker labels are available on paid plans and can be included in the SRT or left out.
- Set the language yourself if the video is short or starts with music. Mixed Chinese and English speech is handled in one transcript.
More from the same MP4 transcript
Once the video is transcribed, the subtitles are only one of the files you can take away. Each export below comes from the same corrected text, so you check the words once.
- VTT for the same video embedded on your own website.
- TXT or DOCX to write the video description, a blog post or show notes.
- CSV if you want to review every cue with its start and end time in a spreadsheet.
- AI prompt to paste the transcript into ChatGPT or Claude and draft chapter titles or a summary yourself.