Converting audios

Converting audio files is a frequently needed task when serving audio files to target services and users.

How to convert large batch of audios ?

Dependencies

Warning : avconv have been removed from Ubuntu packages since 18.04. Use ffmpeg instead.

sudo apt-get install lame ffmpeg
man lame                           # then search for parameters via "/{param}". ex: /-m

Technolect

• cbr: constant bit rate.
• abr: average bit rate.
• vbr: variable bit rate.

For more, see man lame.

Helpers

$ mkdir -p ./new/                    # create folder, if not existing (-p)
$ file="./data/cmn-0a0a8a8b.flac"    # path to .flac file into varible "$file"
$ ffprobe -hide_banner "$file"       # print out metadata of $file, for some formats only

$ ffprobe -hide_banner ./data/cmn-0a0a8a8b.ogg
Input #0, ogg, from './data/cmn-0a0a8a8b.ogg':
  Duration: 00:00:01.25, start: 0.000000, bitrate: 99 kb/s
  Stream #0:0: Audio: vorbis, 44100 Hz, mono, fltp, 80 kb/s
    Metadata:
      TITLE           : 高低
      LICENSE         : Creative Commons BY-SA 3.0 U.S
      COPYRIGHT       : (c) 2009 Yue Tan
      ARTIST          : Tan
      DATE            : 2009-07-08
      GENRE           : Speech
      SWAC_LANG       : cmn
      SWAC_TEXT       : 高低
      SWAC_ALPHAIDX   : gāodī
      SWAC_SPEAK_NAME : Tan
      SWAC_SPEAK_GENDER: F
      SWAC_SPEAK_BIRTH_YEAR: 1978
      SWAC_SPEAK_LANG : zho
      SWAC_SPEAK_LANG_REGION: Liaoning
      SWAC_SPEAK_LIV_COUNTRY: FR
      SWAC_SPEAK_LIV_TOWN: Caen
      SWAC_PRON_PHON  : gāodī
      SWAC_COLL_SECTION: HSK niveau IV
      SWAC_COLL_LICENSE: Creative Commons BY-SA 3.0 U.S
      SWAC_COLL_COPYRIGHT: (c) 2009 Yue Tan
      SWAC_TECH_DATE  : 2009-07-08
      SWAC_TECH_SOFT  : Shtooka Recorder/1.3

Simple batch format conversion

for file in ./flac/*.flac
do
  key=$(basename "$file" .flac).mp3                                 # name of the file minus .flac, plus .mp3 
  lame --abr 24    -m m -h --resample 22.05 "$file" "./new/$key";
done

Metadata-based format conversion

This example works on SWAC recorder audio files available at https://packs.shtooka.net. Those files contain rich metadata, with a SWAC_TEXT metadata field. In this example, we assume a folder with file audio-0a6f36g.flac and metadata SWAC_TEXT : 很.

for file in ./flac/*.flac
do
  key=$(ffmpeg -i "$file" 2>&1 | sed -ne 's/.*SWAC_TEXT *: //p')                 # print metadata, assign SWAC_TEXT's value to variable.
  lame --abr 24    -m m -h --resample 22.05 "$file" "./new-24k/cmn-$key.mp3";    # ex: cmn-很.mp3 (24k abr)
  lame --cbr -b 96 -m m -h --resample 22.05 "$file" "./new-96k/cmn-$key.mp3";    # ex: cmn-很.mp3 (96k cbr)
done

See also

• c:Commons:Audio