Recently I had a need to convert a WAV audio file from 22.05Khz, Mono, 8-Bit PCM to Khz, Mono, 16-Bit PCM so that a phone system I was configuring could interpret the audio files used for on-hold messaging.

There are a number of online services that can do this, but I don’t trust them and I wanted to work out how to do this with ffmpeg.

Since I was using my Windows box at the time and I needed to document this for others, I went with the Windows binary, but the same could be achieved on Linux or Mac.

• The windows package for this software can be found here: https://www.gyan.dev/ffmpeg/builds/ffmpeg-git-essentials.7z
• The download will be a .z7 file which will need to be extracted using 7-zip
• Once extracted, browse to the bin folder, eg. \ffmpeg-2022-06-16-git-5242ede48d-essentials_build\bin
• You will see a file called ffmpeg.exe. This is the file that performs the conversion via command line
• Copy the path from file explorer so that you can browse to this location using command prompt or PowerShell
• Open command prompt, start > run > cmd > enter
• In command prompt, change to this directory , eg. cd C:\ffmpeg-2022-06-16-git-5242ede48d-essentials_build\bin
• Now you will be in the bin directory and you can use ffmpeg
• To check what type of file you are working with, run the following command C:\>ffmpeg.exe -i C:\sample.wav , eg. C:\ffmpeg-2022-06-16-git-5242ede48d-essentials_build\bin>ffmpeg.exe -i C:\users\<user>\downloads\sample.wav
• An output will be given showing the encoding of the file.
• If the file is already 8Khz, Mono, 16-Bit PCM then no changes are necessary and the file can be used.
• If the file is something different to this, for example 22050 Hz, mono, u8, 176 kb/s, then it will need converting.
• Keep in mind that by default ffmpeg will output information in Hz rather than Khz. In this example, 22050 Hz is 22.05 Khz (not what we want!).
• The u8 listed in the example is an ffmpeg audio type. See this page to identify what audio types there are: https://trac.ffmpeg.org/wiki/audio%20types.
• When we check u8 against audio types, it is listed as PCM unsigned 8-bit. Since this is not 16-bit, we need to convert the file.
• The audio type for 8Khz, Mono, 16-Bit PCM is pcm_s16le.
• To convert the file, we use the following command ffmpeg -i sample.wav -acodec pcm_s16le -ac 1 -ar 8000 output.wav, eg: ffmpeg.exe -i C:\users\<user>\downloads\sample.wav -acodec pcm_s16le -ac 1 -ar 8000 C:\users\<user>\downloads\output-16-bit.wav
• Let’s break down the command. -i is for input file, -acodec specifies the audio codec, -ac sets the number of audio channels (mono), -ar sets the audio sampling frequency (8Khz)
• If we run ffmpeg with no switches against the converted sample file again, we get 8000 Hz, mono, s16, 128 kb/s which is compatible with phone system I was working on.

Using the same method above, you could convert any number of source audio formats to other formats. Ffmpeg also has many other capabilities and can convert video too, it’s extremely flexible and robust. I recommend checking out the following man page https://manpages.org/ffmpeg.

Lastly, I looked back through my notes and realised I had used ffmpeg a few times in the past. One of these times was to convert a mp4 video file to webm for a HTML5 project, eg.

ffmpeg -i "C:\Users\<user>\Documents\Media1.mp4" -c:v libvpx-vp9 -crf 30 -b:v 0 -b:a 128k -c:a libopus C:\temp\output.webm

Thanks for reading – Jesse