To repeat a nature ambience video with FFmpeg on Ubuntu, use -stream_loop -1 before the input and, when creating a finite file, set its length with -t. The loop option repeats the source; it does not make a YouTube broadcast, guarantee a seamless join, or create an unlimited output when you specify a duration.
For a one-hour file from a local input, start with the command below. It requests a repeated input, a one-hour output, all input streams and stream copying, subject to compatibility with the source and MP4 container.
Decide whether you need a file or continuous playback
A looped file and ongoing playback solve different problems. If you want a prepared clip that repeats for a known period, render a finite output with -t. You can then inspect that file, move it to a playback machine, or use it in a separate broadcasting workflow.
If you want playback to keep going until you stop it, do not give FFmpeg a finite duration. Instead, arrange an appropriate player or output workflow that consumes the repeating input and has a deliberate way to stop. An ordinary file output without a stopping condition can keep growing, so it is not a sensible substitute for a playback workflow that runs indefinitely.
This distinction matters for an always-on YouTube channel. FFmpeg can prepare or process media, but the command shown here only writes a file; it does not connect to YouTube or send a live stream. If your goal is to broadcast recorded material rather than create a local asset, this guide to streaming recorded language lessons without a VPS discusses a different operational problem. A local loop file may be one part of that setup, not the whole setup.
There is also a second meaning of “continuous”: a loop can run without stopping while still having an audible click, a jump in the image, or a sudden change in the soundscape at each boundary. Repetition handles duration, not the creative edit. Listen to the transition between the end and beginning before using the result for a long ambience session.
Use a bounded FFmpeg loop command
For a local file named input.mp4, this command asks FFmpeg to produce a one-hour MP4 named output.mp4:
ffmpeg -stream_loop -1 -i input.mp4 -t 01:00:00 -map 0 -c copy output.mp4
Replace both filenames with the paths you actually use. Change 01:00:00 to the duration you need, using a time value appropriate to FFmpeg’s duration syntax. The command is a starting point rather than a promise that every source will fit every output container unchanged.
The FFmpeg command-line documentation describes -stream_loop as an input option and documents -1 for infinite repetition; 0 means no loop. Here “infinite” describes how often FFmpeg may repeat the input while processing it. The output is still bounded by -t 01:00:00, so FFmpeg stops writing when the requested output duration is reached.
That is useful when the source is shorter than the planned playback period. A ten-minute forest recording, for example, can be repeated to fill a one-hour render. The resulting file is a new output to check, not a guarantee that the repeated sections join invisibly or that the material is suitable for every player.
If your input is in another directory, give FFmpeg the full path or run the command from the directory containing the file. On Ubuntu, first check that FFmpeg is installed and see which build you have with ffmpeg -version. Ubuntu releases do not all ship the same build; the Ubuntu Jammy FFmpeg manual documents a particular Jammy package, not every current Ubuntu installation. Use man ffmpeg on your machine when an option or encoder name differs from a guide written for another release.
Put the loop option before -i
FFmpeg’s command line is organised around inputs and outputs. Options before an -i apply to the input that follows; options after the input are generally output-side instructions. Because -stream_loop -1 tells FFmpeg how to read and repeat the input, place it before the corresponding -i:
ffmpeg -stream_loop -1 -i input.mp4 ...
In this example, the loop setting applies to input.mp4. Putting it after -i changes the point in the command where FFmpeg sees the option and can make it apply in the wrong context rather than setting up repeated input reading. If you have more than one input, take care to associate each input option with the intended -i rather than treating the command as an unordered collection of switches.
The same input/output distinction is useful when adapting examples. -i input.mp4 identifies the source. -t, -map and -c in the example are output requests placed after the input. The output filename at the end gives FFmpeg somewhere to write the result. Changing the order casually can change what an option means, so when a command behaves unexpectedly, compare its layout with the installed manual rather than moving flags at random.
If you are instead trying to make a live broadcast run continuously from a machine, there are additional questions: how the output is sent, how it is monitored, and what happens after a process exits. The Linux server guide for recorded church services concerns that broader workflow. Keep the distinction clear: the loop command above creates a local output file and has no YouTube destination.
Read the duration, mapping and copy requests
-t 01:00:00 asks FFmpeg to stop the output after the specified duration. It is what makes the example a finite render despite the infinitely repeated input. It does not tell FFmpeg to create an unlimited file. If you omit -t, you have removed this duration limit, not added a sensible stopping rule for a file-based workflow.
-map 0 requests all streams from input number zero. A video file may contain more than a picture and a single audio track: it could include several audio tracks, subtitles, or other stream types. Mapping all streams is explicit, but the output container and stream types still have to be compatible. If you only want particular streams, review FFmpeg’s mapping documentation and make a deliberate selection rather than assuming that every embedded track belongs in the output.
-c copy asks FFmpeg to copy encoded streams rather than decode and re-encode them. When the streams and target container are suitable, that avoids another lossy encoding pass and is generally less work than transcoding. It does not convert a codec into a different codec, repair a source, or ensure that any arbitrary combination of streams can be written into MP4.
That trade-off is often useful for ambience footage already in a format that your intended output can carry. But if FFmpeg reports that a stream cannot be muxed into the requested container, or your player cannot handle the result, remove the copy request and choose compatible encoders and settings. Re-encoding takes more processing and may change quality; it gives you a route to a different codec or stream layout when simple copying is unsuitable.
These options are requests, not proof that the resulting file is correct. Check FFmpeg’s completion status and inspect the output itself. If the file contains tracks you did not expect, revisit mapping. If it fails at the container boundary, investigate stream compatibility rather than assuming that the loop setting is at fault.
Check container and stream compatibility
An .mp4 filename alone does not make all stream types valid MP4 contents. The input might contain a video codec, audio codec, subtitle stream or metadata arrangement that the target container or your playback device does not accept. With -map 0 -c copy, you have asked to keep every input stream and copy it; that is convenient only when those streams can be carried through as requested.
If the output fails, simplify the diagnosis. First try mapping only the video and audio streams you need, using explicit stream selection after checking what the source contains. Then test whether the intended container accepts those streams. If not, choose encoders supported by your installed FFmpeg build and by the playback workflow. The FFmpeg project’s documentation index points to the command-line and format documentation; consult it alongside your local manual because build options can vary.
Stream copying and re-encoding are not competing answers for every case. Copying preserves the existing encoded stream and avoids another lossy encode, but it cannot change the codec. Re-encoding can make a stream suitable for a different container or device, at the cost of encoding time and a generation of quality loss if a lossy codec is used. For a long ambience file, make a short test first and check the result in the actual player before committing to a lengthy render.
If your video has multiple segments rather than one source file, do not assume that placing files one after another is equivalent to looping one input. Concatenation workflows depend on compatible stream properties or prior preparation. The FFmpeg FAQ on concatenating media covers approaches, but it is not a claim that arbitrary clips can be joined unchanged. When the boundary itself needs to disappear, edit the source or use a suitable transition workflow and inspect the result.
Test the output in the intended playback workflow
A successful FFmpeg exit is only one check. Open the output in the player, broadcast software or other workflow that will actually use it. Confirm that the image displays, the sound is present, the file seeks as expected, and the duration is what you intended. If you have multiple tracks, verify that the desired language or audio mix is selected rather than trusting the default.
Pay particular attention to the loop boundary. Listen near the end and start of adjacent repetitions, and watch for a change in framing, exposure, motion or colour that becomes distracting. Nature recordings often include wind, birdsong or water with changing levels; a small discontinuity can become conspicuous when repeated. The loop option repeats the source, but it does not crossfade the end into the beginning or repair mismatched audio and video lengths.
If the cut is abrupt, you have several practical choices. Choose a different in/out point in the source, edit the boundary, or create a transition with an editing workflow that suits the media. There is no universal seamless-loop command for every scene and sound. If you need to re-render after an edit, repeat the same compatibility check and test the new output in the target player.
Also test the operational hand-off. A finite file can be queued or replayed by a separate playback tool, but that tool controls its own behaviour when the file ends, loses access to media, or encounters an unsupported track. For a channel that needs an unattended overnight run, check restart and monitoring behaviour separately. The guide to automatically restarting an FFmpeg YouTube stream with systemd addresses process recovery, which is distinct from making this looped file.
If the repeated file is ready but the concern is keeping a broadcast running while your own computer is off, that is a separate operating burden. StreamNeo removes the need to keep your local computer running for this particular always-on playback job: upload the prepared video and provide your YouTube stream key, then the broadcast is monitored and restarted if it drops. It is YouTube-only, so it is not a replacement for FFmpeg when you need local conversion or editing.
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
Does -stream_loop -1 make the output last forever?
It repeats the input indefinitely while FFmpeg processes it, but the example includes -t 01:00:00, which bounds the output to one hour. Remove the duration only if you have a deliberate continuous playback or output workflow and a suitable stopping method; an unbounded file can keep growing.
Why is -stream_loop -1 before -i?
It is an input option, so it belongs before the -i for the file it should repeat. FFmpeg distinguishes input-side and output-side options, and their position in the command matters.
Can I leave -c copy in every loop command?
No. It asks FFmpeg to copy streams without re-encoding, but the source streams and output container must be compatible. If copying fails or the target player rejects the file, select compatible streams or re-encode with suitable codecs.
Does this command send my ambience video to YouTube?
No. It writes a local output file; it contains no YouTube destination or live-streaming setup. Use a separate playback and broadcast workflow if your goal is a YouTube live channel, and test that workflow independently.