How to convert a sound file to Opus

For testing, I want to convert an MP3 and WAV file I have to Opus, what are the steps to doing this?


In newer Ubuntu releases the Opus codec is included in the libavcodec libraries that will be installed with ffmpeg. Audio encoding is then done with

ffmpeg -i infile.ext <options> outfile.opus

The audio converter shipped with the opus-tools can convert audio in raw, wave or AIFF format. The minimal syntax uses default settings:

opusenc input.wav output.opus

We may want to add a better bitrate as the default 96 kbps with the option --bitrate N.nnn (for all options consult the manpage for opusenc).

To convert mp3 "on the fly". i.e. without creating a temporary file we can pipe the output from ffmpeg to opusenc like this:

ffmpeg -i input.mp3 -f wav - | opusenc --bitrate 256 - output.opus

Ubuntu 14.04 and Debian 8 ship with version 9 of libav-tools in their repositories, and it has built-in support for Opus through the package libopus0.

Example 1: Reencode an audio file as opus

With version 9 of libav-tools and libopus0 installed you can simply, for example, do:

avconv -i file.mp3 -map 0:a -codec:a opus -b:a 100k -vbr on file.opus

What the options do

  • -i file.mp3 sets the input file.
  • -map 0:a will select all audio streams (a) from the input file 0. Read more about -map on https://libav.org/avconv.html#Advanced-options
  • -codec:a opus selects the opus encoder for the audio streams (a). Read more about -codec on https://libav.org/avconv.html#Main-options.
  • -b:a 100k sets the audio's bitrate to 100 kilobit/s. Read more about -b on https://libav.org/avconv.html#Codec-AVOptions
  • -vbr on turns on variable bitrate. This is an option specific for libopus. Here are all options for libopus:

    $ avconv -h full | grep opus -A 11
    avconv version 9.11-6:9.11-3+b2, Copyright (c) 2000-2013 the Libav developers
      built on Apr  6 2014 17:45:45 with gcc 4.8 (Debian 4.8.2-16)
    libopus AVOptions:
    -application       <int>   E..A. Intended application type
       voip                    E..A. Favor improved speech intelligibility
       audio                   E..A. Favor faithfulness to the input
       lowdelay                E..A. Restrict to only the lowest delay modes
    -frame_duration    <float> E..A. Duration of a frame in milliseconds
    -packet_loss       <int>   E..A. Expected packet loss percentage
    -vbr               <int>   E..A. Variable bit rate mode
       off                     E..A. Use constant bit rate
       on                      E..A. Use variable bit rate
       constrained             E..A. Use constrained VBR
    
  • file.opus sets the output file.

Example 2: Grab the audio from a video file and encode it as opus

Take the second stream of the first input (-map 0:1), which is the audio stream. Encode it with libopus at 100 kbit/s with variable bitrate on:

$ avconv -stats -i linuxactionshowep309-432p.mp4 -map 0:1 -c libopus -b 100k linuxactionshowep309-432p-audio-only.opus
avconv version 9.11-6:9.11-3+b2, Copyright (c) 2000-2013 the Libav developers
  built on Apr  6 2014 17:45:45 with gcc 4.8 (Debian 4.8.2-16)
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'linuxactionshowep309-432p.mp4':
  Metadata:
    major_brand     : isom
    minor_version   : 512
    compatible_brands: isomiso2avc1mp41
    encoder         : Lavf55.33.100
  Duration: 01:14:48.45, start: 0.042667, bitrate: 466 kb/s
    Stream #0.0(und): Video: h264 (High), yuv420p, 768x432 [PAR 1:1 DAR 16:9], 330 kb/s, 30 fps, 30 tbr, 15360 tbn, 60 tbc
    Stream #0.1(und): Audio: aac, 48000 Hz, stereo, fltp, 128 kb/s
Output #0, ogg, to 'linuxactionshowep309-432p-audio-only.opus':
  Metadata:
    major_brand     : isom
    minor_version   : 512
    compatible_brands: isomiso2avc1mp41
    encoder         : Lavf54.20.3
    Stream #0.0(und): Audio: libopus, 48000 Hz, stereo, flt, 100 kb/s
Stream mapping:
  Stream #0:1 -> #0:0 (aac -> libopus)
Press ctrl-c to stop encoding
size=   54360kB time=4488.47 bitrate=  99.2kbits/s    
video:0kB audio:53875kB global headers:0kB muxing overhead 0.900602%

With the package mediainfo installed:

$ mediainfo linuxactionshowep309-432p-audio-only.opus
General
Complete name                            : linuxactionshowep309-432p-audio-only.opus
Format                                   : OGG
File size                                : 53.1 MiB
Duration                                 : 1h 14mn
Overall bit rate                         : 99.2 Kbps
Writing application                      : Lavf54.20.3
major_brand                              : isom
minor_version                            : 512
compatible_brands                        : isomiso2avc1mp41

Audio
ID                                       : 2104437746 (0x7D6F2BF2)
Format                                   : Opus
Duration                                 : 1h 14mn
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 48.0 KHz
Compression mode                         : Lossy
Writing library                          : Lavf54.20.3

Opus on 12.04

On 12.04 (Precise), however, there are dependency problems with installing the opus codecs and tools, so I have found by far the best solution is the one that has become available very recently: compile the opus audio encoder and decoder as noted here, and build ffmpeg with opus support by adding --enable-opus to the configure options of ffmpeg (as listed on the compilation guide).

I know that ffmpeg is deprecated in Ubuntu in favour of Libav, but compiling is a good way to get a fully functioning opus encoder/decoder integrated into ffmpeg itself. You can then use it to convert files (first to wav) and then to .opus. The documentation installed with libopus and ffmpeg will reveal all the options that can be used to convert files.

When converting files with ffmpeg after compilation, you must specify -acodec libopus or ffmpeg will not use the opus codec:

ffmpeg -i pc.wav -ar 48000 -ac 2 -acodec libopus -ab 256k man.opus

You can then test the file created with

ffplay man.opus

Compilation Tips

There's no need to reproduce the guide here in its entirety, but it's worth noting one or two things:

  • You should first install the dependencies as listed (I omit yasm from the list: see my second point):

     sudo apt-get -y install autoconf build-essential checkinstall git libass-dev libfaac-dev libgpac-dev libjack-jackd2-dev libmp3lame-dev libopencore-amrnb-dev libopencore-amrwb-dev librtmp-dev libsdl1.2-dev libtheora-dev libtool libva-dev libvdpau-dev libvorbis-dev libx11-dev libxext-dev libxfixes-dev pkg-config texi2html zlib1g-dev
    
  • There is one issue that should be pointed out: the git build seems to want yasm-1.2, and that is not available, so you have to compile the source from the official site, but it is simple. Just remove any installed versions of yasm, then unpack the downloaded archive, cd to the folder, run ./configure && make and then sudo checkinstall. If any other builds require the earlier version, you can just remove this version and install the repository version.

  • It is necessary to remove any existing libav, ffmpeg, x264, libvpx, or fdk-aac packages before you start compiling.

  • It is critical that you compile and install x264, fdk-aac, libvpx and opus before you build ffmpeg, as those libraries will be used in the build.

  • Do not forget to add --enable-opus to the configure options when you run the ffmpeg compilation.

  • The version of opus compiled was 1.1alpha, so you may need to re-compile the opus library and ffmpeg in the future again when a new version is released.

  • You can use ffplay to play any opus files you create.


That's how I do it:

  • First, open a terminal in the same directory where your audio files are.
  • Then, type this command:
$ opusenc --bitrate 320 --max-delay 10 "18 - Soul Asylum - Runaway Train (Album Version).flac" "18 - Soul Asylum - Runaway Train (Album Version).opus"

EDIT:

For Audiophiles:

$ opusenc --bitrate 510 --max-delay 10 "18 - Soul Asylum - Runaway Train (Album Version).flac" "18 - Soul Asylum - Runaway Train (Album Version).opus"

No need to specify --maxdelay 10 option because opusenc do this by default.

Console Output for this file conversion (--bitrate 320):

    Encoding using libopus 1.1.2 (audio)
    -----------------------------------------------------
       Input: 44.1kHz 2 channels
      Output: 2 channels (2 coupled)
          20ms packets, 320kbit/sec VBR
     Preskip: 356

    Encoding complete   
    -----------------------------------------------------
           Encoded: 4 minutes and 22.4 seconds
           Runtime: 8 seconds
                    (32.8x realtime)
             Wrote: 10955530 bytes, 13120 packets, 13124 pages
           Bitrate: 317.691kbit/s (without overhead)
     Instant rates: 1.2kbit/s to 510.4kbit/s
                    (3 to 1276 bytes per packet)
          Overhead: 4.89% (container+metadata)

It's super fast! Less than 8 seconds with a complexity of 10 (Encoding computational complexity (0-10, default: 10). Zero gives the fastest encodes but lower quality, while 10 gives the highest quality but slower encoding) and a maximum delay time of 10ms (Maximum container delay in milliseconds (0-1000, default: 1000)), so if you skip time in a song, the clipping effect will have a duration of 10ms so it is inperceptible (try with 1000 and hear the difference skipping time with your mouse). Bitrate is VBR by default. 320kbps worked for me so is optional, play with this number:
--bitrate N.nnn => Target bitrate in kbit/sec (6-256 per channel)

By the way, encoding from MP3 to OPUS is not a good idea, it's not going to sound better, their compression algorithms are way too different. But from FLAC or WAV or any other Lossless Audio Format, that's another story.

Note: To encode another file, just press the Up Arrow in the same terminal to call the last command and change the name of the input and output files.
If you are looking for a ffmpeg/avconv GUI, maybe TraGtor is what you need.

You can also check the spectogram differences between Lossless and Lossy formats at high bitrates with Spek or Audacity.