Published 2022-01-28. Last modified 2023-11-28.
Time to read: 3 minutes.

This page is part of the av_studio collection.

media_trim

This article discusses the technical implementation of MediaTrim, which I presented in the previous article.

Codecs and Re-Encoding

• Some of the ffmpeg options this script uses are not available in older versions of ffmpeg.
• The most important thing to know about options that might be passed to ffmpeg is that arbitrary start and end times can be specified accurately when trimming if the video stream is re-encoded. To force a video re-encoding, simply do not specify the -vcodec copy option.
• Non-essential streams in video files are discarded by the MediaTrim trimming process. See Selecting streams with the -map option.

Sony A7iii Media Files

I wanted to trim a video file created by my Sony A7iii camera, so only the portion from 0:51 through 2:45 was extracted.

FFprobe Provides Useful Information

FFprobe gathers information from multimedia streams and prints it in human- and machine-readable fashion.

For example, it can be used to check the format of the container used by a multimedia stream and the format and type of each media stream contained in it.

 – From FFprobe documentation

Let’s use ffprobe to examine the streams within the media file.

Shell

$ ffprobe myvideo.mp4
ffprobe version 5.1.2-3ubuntu1 Copyright (c) 2007-2022 the FFmpeg developers
  built with gcc 12 (Ubuntu 12.2.0-14ubuntu2)
  configuration: --prefix=/usr --extra-version=3ubuntu1 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --arch=amd64 --enable-gpl --disable-stripping --enable-gnutls --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libcodec2 --enable-libdav1d --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libglslang --enable-libgme --enable-libgsm --enable-libjack --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librabbitmq --enable-librist --enable-librubberband --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libssh --enable-libsvtav1 --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzimg --enable-libzmq --enable-libzvbi --enable-lv2 --enable-omx --enable-openal --enable-opencl --enable-opengl --enable-sdl2 --disable-sndio --enable-libjxl --enable-pocketsphinx --enable-librsvg --enable-libmfx --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libx264 --enable-libplacebo --enable-librav1e --enable-shared
  libavutil      57. 28.100 / 57. 28.100
  libavcodec     59. 37.100 / 59. 37.100
  libavformat    59. 27.100 / 59. 27.100
  libavdevice    59.  7.100 / 59.  7.100
  libavfilter     8. 44.100 /  8. 44.100
  libswscale      6.  7.100 /  6.  7.100
  libswresample   4.  7.100 /  4.  7.100
  libpostproc    56.  6.100 / 56.  6.100
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x560a973df5c0] st: 0 edit list: 1 Missing key frame while searching for timestamp: 1001
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x560a973df5c0] st: 0 edit list 1 Cannot find an index entry before timestamp: 1001.
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'myvideo.mp4':
  Metadata:
    major_brand     : XAVC
    minor_version   : 16785407
    compatible_brands: XAVCmp42iso2
    creation_time   : 2023-01-05T00:52:24.000000Z
  Duration: 00:10:58.16, start: 0.000000, bitrate: 51445 kb/s
  Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709/bt709/iec61966-2-4, progressive), 1920x1080 [SAR 1:1 DAR 16:9], 49370 kb/s, 59.94 fps, 59.94 tbr, 60k tbn (default)
    Metadata:
      creation_time   : 2023-01-05T00:52:24.000000Z
      handler_name    : Video Media Handler
      vendor_id       : [0][0][0][0]
      encoder         : AVC Coding
  Stream #0:1[0x2](und): Audio: pcm_s16be (twos / 0x736F7774), 48000 Hz, 2 channels, s16, 1536 kb/s (default)
    Metadata:
      creation_time   : 2023-01-05T00:52:24.000000Z
      handler_name    : Sound Media Handler
      vendor_id       : [0][0][0][0]
  Stream #0:2[0x3](und): Data: none (rtmd / 0x646D7472), 491 kb/s (default)
    Metadata:
      creation_time   : 2023-01-05T00:52:24.000000Z
      handler_name    : Timed Metadata Media Handler
      timecode        : 03:52:30:26
Unsupported codec with id 0 for input stream 2 

FFprobe shows us that myvideo.mp4 has 3 streams, numbered from origin zero:

0. MPEG-4 video stream: encoded using the H.264 high profile, which is the most commonly used H.264 profile, with 1920x1080 resolution, recorded at 50 Mb/s, and 59.94 fps. This stream was recoded to the same format using FFmpeg’s default stream handling. The recoding process computes keyframes for the new start and points after trimming.
1. CD-quality audio stream: pcm_s16be (16-bit WAV format), at 48 kHz, in stereo, with a bit rate of 1536 Kb/s. In contrast, 24 bits are commonly used in 2023 for streaming audio. WAV encoding is uncompressed, and so it is disallowed by strict MP4 compliance.
   
   Because FFmpeg follows string mp4 compliance by default, FFmpeg will not copy audio encoded in WAV format. However, the audio stream can be transcoded to a compressed format, so it can be included in the output file. You could relax FFmpeg’s strict mp4 compliance by providing the -strict experimental option to FFmpeg. Instead, I decided to compress the audio using AAC because this results in a smaller file that sounds just as good.
2. Data stream: normally ignored by ffmpeg when creating output. ffprobe shows an error for this stream (Unsupported codec with id 0 for input stream 2), but we do not need this stream, so the trim script will also ignore it.

Transcoding Audio

Audio is best transcoded to a lossless compressed format that is supported by the MP4 container, like aac or Opus.

The hierarchy of audio encoder quality has been reported as: libopus > libvorbis >= libfdk_aac > libmp3lame >= eac3/ac3 > aac > libtwolame > vorbis > mp2 > wmav2/wmav1.

Ffmpeg’s AAC transcoder gives poor results for bit rates less than 128k. Opus gives a better result; however, some hardware devices may not have support. My 3-year-old TCL TV does not support Opus, for example.

Ffmpeg supports five AAC encoders, including the regular AAC encoder and the Fraunhofer FDK AAC codec. The latter requires that you compile FFmpeg because of licensing issues. This article will not get into that.

Regular AAC codec

The following is the incantation used by the trim script. It yields a regular mp4 with a video stream and an AAC stereo stream.

Shell

$ ffmpeg -y -ss 51 -to 2:45 -i input.mp4 -acodec aac output.mp4

Both the video and audio streams are recoded. The video stream selected from the input file is the highest resolution video stream in that file. Non-essential streams for regular playback are ignored.

😁

The resulting file is a balance of the highest-quality audio and video streams with the smallest overall file size that can usually be expected to play on most devices.

Fraunhofer FDK AAC codec

You could go one better and build FFmpeg with the Fraunhofer FDK AAC codec, then invoke it with:

Shell

$ ffmpeg -y -ss 51 -to 2:45 -i input.mp4 -c:a libfdk_aac output.mp4

Share to Twitter, Hacker News, LinkedIn.