Burn subtitles into each prerecorded video before you send it into a continuous YouTube stream. The workflow has two distinct stages: render timed text onto the video frames, then play those prepared files through a scene or playlist that feeds your stream.
That is different from uploading captions in YouTube Studio or adding a selectable subtitle track. Burned-in text is always part of the picture; it cannot be switched off by a viewer. Keep your clean originals and subtitle files so you can revise the wording, timing or appearance later.
Burned-in text or a subtitle track?
A subtitle file such as SRT contains text paired with timings. By itself, it does not alter the video. A player or stream setup may read a separate subtitle track and let viewers select it, but that depends on the format, delivery path and player support.
Burn-in, also called hard subtitles, is a different operation: software draws each timed line onto the video frames and encodes a new video. The text then appears for everyone watching that prepared file, whether or not they have subtitle controls. You cannot subsequently hide it or restyle it at the viewer’s end.
| Approach | What appears to the viewer | Practical trade-off |
|---|---|---|
| Burned into the picture | Text is visible as part of the video image | The video must be processed; viewers cannot switch off or restyle the text |
| Separate subtitle track | A player may offer a selectable caption track | Depends on the container, stream path and player supporting that track |
| YouTube Studio captions | Captions are associated with a video in YouTube’s caption workflow | This is not the same as rendering words into the frames of a separate continuous live feed |
YouTube Help documents adding subtitles and captions to videos through upload, manual entry and auto-sync. Those controls are useful when publishing a video with captions, but they do not mean that an uploaded sidecar file becomes part of the picture in your continuous live output. Check YouTube’s current subtitle and caption guidance for its supported workflows and current conditions.
The useful choice depends on what your audience needs. If a devotional loop must show lyrics to every viewer without relying on a player setting, burn them in. If viewers need to select languages or turn text off, investigate a supported separate-track route for your actual playback and delivery chain instead; do not assume a track will work in every live setup. FFmpeg’s documentation includes a separate WebVTT rendition example for HLS, but that describes an HLS method, not automatic compatibility with every YouTube Live ingest path.
Prepare timed subtitle files
Start with a clean copy of the video and a subtitle file whose timings match that exact clip. SRT and ASS are common inputs for the FFmpeg subtitle workflow described below. A subtitle file is not just a transcript: each line needs a start and end time so the text appears while the corresponding speech, lyric or on-screen moment occurs.
If you have words but no timings, make or synchronise captions before rendering. YouTube’s caption guidance describes upload and auto-sync options for videos, but auto-sync has conditions: it depends on supported speech-recognition languages, and YouTube does not recommend it for videos longer than an hour or with poor audio. Those notes concern YouTube’s own workflow; they are not a guarantee that its resulting file will suit every editing or streaming job.
Name files so each subtitle file is clearly paired with its video. For example, morning-bhajan.mp4 and morning-bhajan.srt are less likely to be confused with another clip than generic names such as video.mp4 and captions.srt. Keep a separate working folder for originals, subtitle sources, and rendered outputs. This makes it easier to find which version needs correction when you discover a typo during review.
Before processing a full collection, inspect the timing near the beginning and end of each file. A clip may have an opening title, a long instrumental passage, or a closing pause that makes a transcript-only check misleading. If the subtitles drift against the audio, correct the timings first; styling cannot repair synchronisation.
Also decide how much of the image the text may cover. Lyrics near a singer’s face, a temple image, a news ticker or a business logo may obscure information. A consistent position and readable contrast are more useful than elaborate styling. If you want captions in more than one language, retain each language’s source file and decide whether you are preparing separate versions or using another supported delivery method.
Render subtitles into each video with FFmpeg
FFmpeg’s official documentation describes the subtitles video filter for rendering subtitle text into a video. A basic example, also shown in a current technical tutorial, is:
ffmpeg -i input.mp4 -vf subtitles=subtitle.srt output.mp4
Replace the example filenames with your own paths. The input is the original video, the filter reads the timed subtitle file and draws its text over the frames, and the output is a newly encoded video. Read the FFmpeg filter documentation for the filter’s current syntax and requirements before adapting the example to your system.
This command is a starting point, not a claim that it has been tested on your machine. The subtitle filter relies on build support for its subtitle rendering dependencies; if FFmpeg reports that the filter is unavailable, check how your installed build was compiled rather than assuming the subtitle file is faulty. A technical FFmpeg subtitles tutorial demonstrates SRT and ASS use, styling through force_style, and font directory options for ASS input. Treat its examples as syntax guidance and verify details against your own installed version.
Burn-in changes the picture, so the video portion has to be processed and encoded; it cannot simply copy the original video stream unchanged. Audio handling and stream mapping depend on the source and the output you need. Avoid copying a long command from a different file and assuming it will suit yours: container, codecs, audio streams and subtitle filter availability can vary. First render a short representative clip, then inspect the output and its audio before choosing settings for a larger batch.
For styled ASS subtitles, review fonts and style fields as well as position and outline. A font name in a subtitle file does not ensure that the font is installed where the render runs. FFmpeg’s filter documentation covers fontsdir for directing the renderer to fonts; test on the actual machine that will prepare your files, particularly if you move the job between computers.
Keep the command and a note of the input/output names alongside your project files. If you need to change line breaks, spelling or the safe-area position, the source video and subtitle file let you render again without degrading an already rendered copy. Store the finished output under a distinct name so it is not mistaken for the clean original.
Check the text, timing and output
Watch the whole test render, not only the first subtitle. Check the opening, middle and ending, and include points where the background becomes bright or visually detailed. A subtitle that is clear against a dark shot can disappear against a white sky or moving pattern. Look for lines that appear too early, linger after speech, overlap, or sit so low that a viewer’s interface may cover them.
Confirm the text is legible at the size in which people are likely to watch. Long lines may be hard to read on a phone, while a heavy outline or opaque box can cover artwork. If you adjust font, size, outline or position, render a new short sample and review it. The appropriate style depends on the material: a lofi station with changing scenery has different contrast challenges from a fixed local-news panel.
Check that the output opens and plays from beginning to end, that the audio is present and synchronised, and that its dimensions and overall appearance are what you intend to send to the playback scene. FFmpeg’s filter can render text, but the source material and chosen output settings affect the final result. Keep an untouched original and the timed caption file; neither a burn-in nor a later correction should be the only surviving copy.
If your library contains many videos, review a sample from each type of content before rendering everything. A lyric format that works on one bhajan may not fit another clip with a different aspect ratio. Create a small checklist for spelling, timing, screen position, sound and playback, then use it consistently. This is especially helpful when another person prepares files for the channel.
A poor stream picture is not always a subtitle problem. If the rendered clip itself is clear but the live output looks soft, compare the source and stream path separately; our guide to common fixes for a blurry YouTube live stream helps distinguish source quality from stream settings. Do not compensate for a weak source by increasing unrelated settings without checking the actual file first.
Play prepared videos through OBS
Once a video has its subtitles burned in and has passed review, add the prepared file to the playback stage. OBS Media Sources can play video files and document controls for playlist order and shuffle. A media source is a way to put a file into an OBS scene; it is not the step that creates burned subtitles. The text is already in the prepared video.
The OBS Project’s Media Sources guide explains file playback and playlist controls. Build a simple test scene with one prepared clip first. Confirm that it starts, reaches the end, and behaves as expected before you add a larger queue. OBS also documents subtitle-track display settings for suitable sources, but that is a separate-track feature and is not needed for the burned-in approach.
For a sequence, add the files in the order you want viewers to see them and review the playlist behaviour in your own OBS version. Decide what should happen when a clip ends and whether shuffle is appropriate. A scheduled news loop may need a deliberate order; an ambience channel might accept repetition or a varied sequence. The important distinction is that the prepared videos themselves carry visible text, while OBS is responsible for presenting the files through the scene.
OBS output then needs to connect to the YouTube Live setup you have configured for your account. YouTube’s ingest instructions and the settings available to your account are the authority for that connection. This guide does not prescribe one ingest protocol, stream key arrangement, latency mode or recovery plan: those depend on your actual setup. If you use a cloud workflow to avoid leaving a home computer running, StreamNeo can take the prepared video file and run it as a continuous YouTube broadcast, removing the need to keep your own computer on for playback.
Build a continuous playback workflow
Think of the work as a small production line: retain the original, prepare and review the timed subtitle file, render a new subtitled video, and place only approved outputs into the continuous playback queue. Keep a clear distinction between “ready to render” and “ready to stream”. It prevents an unfinished subtitle file or an unreviewed render from reaching the live scene.
Use a naming convention that shows the content and version, and avoid overwriting originals. A simple folder structure can separate source, captions, rendered and approved files. The exact arrangement is yours; what matters is being able to match each output to the subtitle source that produced it. If a lyric is corrected, update the caption source, create a revised render and replace the approved file deliberately.
For a channel that cycles through a modest collection, a single ordered playlist may be enough. For programmes that change by time of day, use separate scenes or queues and test each transition. OBS playlist order and shuffle features are playback controls; they do not synchronise subtitle timing across different clips. Each subtitle file must match its own video before it enters the sequence.
Think about audio continuity as well as the picture. A queue can move from one clip to another successfully while leaving a noticeable pause, level change or mismatch in sound. Review the joins with headphones or speakers at a comfortable level. For a music-led Hindi stream, our practical notes on avoiding audio gaps between songs cover another part of the hand-off problem; subtitle rendering does not solve transitions in audio.
Keep a record of which file is in the active queue and where the clean source lives. If a viewer reports a mistimed line, you can identify the affected clip rather than replacing an entire library. For channels with frequent updates, make a small test scene or non-live review step part of the routine before promoting revised files to the always-on sequence.
Test the stream before going live
A correct render is only one part of the result. Test the actual playback scene and the YouTube stream configuration you intend to use, with a private, unlisted or otherwise suitable test where available. Check what the viewer receives, not only the OBS preview: confirm that the text is visible, the audio is present, the sequence changes as intended and the picture remains readable after it reaches YouTube.
Include a clip with captions near its start and one with text near its end. Let the queue cross a transition, and observe whether the next video begins with the expected image and sound. If you plan to loop or shuffle, test that specific behaviour. Do not assume a setting used for a single clip will behave identically in a long-running sequence.
Write down the settings and account-specific connection details that work for your setup, but keep credentials private. If the connection fails, separate the diagnosis: first confirm local playback and scene output, then check the configured YouTube Live connection and current account instructions. For network symptoms, use a measured diagnosis rather than changing bitrate on guesswork; our guide to testing for packet loss before changing bitrate covers that check.
There is no universal configuration established for every computer, OBS version, file library and YouTube ingest path. A test can expose issues before viewers depend on the channel, but it cannot guarantee that future playback or connection conditions will be identical. Keep the source assets and a known-good prepared file available so you can correct a caption or replace a damaged output without rebuilding from memory.
When the file, queue and account-specific stream connection have been checked, decide how you will make updates and recover from a stopped broadcast. Keep that plan separate from the subtitle workflow: burned text makes captions visible in each prepared image, while a reliable continuous stream also depends on playback and the configured live connection.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Can viewers turn off burned-in subtitles?
No. Burned-in text is part of the video image, so viewers cannot switch it off or select a different appearance through a subtitle control. If viewer choice matters, investigate a supported separate-track approach for the specific playback and stream path.
Does uploading an SRT in YouTube Studio burn it into my live stream?
No. YouTube Studio’s subtitle workflow attaches captions to a video; it is not the same as rendering timed text onto frames in a prerecorded file used in a continuous live output. For visible text in that file, render a prepared copy before playback.
Do I need to re-encode the video?
Burn-in changes the video frames, so the video has to be processed and encoded into an output file. Audio handling depends on your source and output requirements, so test a short clip and check both picture and sound before processing a collection.
Can I use one FFmpeg command for every video?
The basic subtitle-filter example is a useful starting point, but files, subtitle styles, fonts and installed FFmpeg builds differ. Render and review a representative clip, then adapt the command to your inputs rather than assuming one command or stream setup suits every channel.