Overview of inputs
input_0.mp4
has the desired video stream and input_1.mp4
has the desired audio stream:

In ffmpeg
the streams look like this:
$ ffmpeg -i input_0.mp4 -i input_1.mp4
Input
Duration: 00:01:48.50, start: 0.000000, bitrate: 4144 kb/s
Stream
Stream
Input
Duration: 00:00:30.05, start: 0.000000, bitrate: 1754 kb/s
Stream
Stream
ID numbers
ffmpeg
refers to input files and streams with index numbers. The format is input_file_id:input_stream_id
. Since ffmpeg
starts counting from 0, stream 1:1
refers to the audio from input_1.mp4
.
Stream specifiers
This can be enhanced with stream specifiers. For example, you can tell ffmpeg
that you want the first video stream from the first input (0:v:0
), and the first audio stream from the second input (1:a:0
). I prefer this method because it's more efficient. Also, it is less prone to accidental mapping because 1:1
can refer to any type of stream, while 2:v:3
only refers to the fourth video stream of the third input file.
Examples
The -map
option instructs ffmpeg
what streams you want. To copy the video from input_0.mp4
and audio from input_1.mp4
:
$ ffmpeg -i input_0.mp4 -i input_1.mp4 -c copy -map 0:0 -map 1:1 -shortest out.mp4
This next example will do the same thing:
$ ffmpeg -i input_0.mp4 -i input_1.mp4 -c copy -map 0:v:0 -map 1:a:0 -shortest out.mp4
-map 0:v:0
can be translated as: from the first input (0
), select video stream type (v
), first video stream (0
)
-map 1:a:0
can be translated as: from the second input (1
), select audio stream type (a
), first audio stream (0
)
Additional Notes