? normalizing audio volume using new ffmpeg options

[hydra3333]

I gather that ffmpeg "-vol" option is deprecated and ffmepg has new audio options, eg http://ffmpeg.mplayerhq.hu/ffmpeg-filters.html#volume

• volumedetect
• volume=volume=+3.1dB

but neither googling nor the ffmpeg documentation show me how to use these new options (in windows) to normalize audio volume.

I hope someone can show me how to do something broadly along the lines:

ffmpeg -i a.mpg [audio parsing options to establish normalization value]
ffmpeg -i a.mpg [video conversion options] [audio conversion options] -af "volume=+[?normalization increase?]dB" out.mp4

Sample scripts and advice would be most welcome.

... if worst comes to worst, then I guess I will need to extract the audio and normalize in separate steps, similar to
http://superuser.com/questions/323119/how-can-i-normalize-audio-using-ffmpeg
and
http://1024.gr/blog/normalize-audio-movies-using-ffmpeg-and-sox-windows
and use something this to recode/mux the video and normalized audio at the same time:

ffmpeg -i a.mpg -i a_normalized.wav -map 0:v -map 1:a [video conversion options] [audio conversion options] out.mp4

(I gather -map 0:v excludes the audio from the first input file, since we need to discard that un-normalized audio ?)


PS also, is this still true of the current version of ffmpeg ? :
http://ffmpeg.org/trac/ffmpeg/ticket/720

-af was never integrated in the main repository,
and using -vf for audio filters i evidently wrong,
so the only way to use audio filtering is through the lavfi input device.
You use it this way:
ffmpeg -f lavfi -i "amovie=INFILE,volume=0.1" OUTFILE

edit: https://forum.videohelp.com/threads/352047-Increasing-audio-volume-on-an-MPG?p=2209366&...=1#post2209366 show -af works OK

PPS if anyone knows how to compress the audio range at the same time, some hints would be handy

[Selur]

1. run volumedetect:

ffmpeg -i "h:\TestClips&Co\test.avi" -af volumedetect -f null -

and decide how much you want to boost the audio.
2. process audio & video (assuming your source only has one audio&video stream)

ffmpeg -i "path to input file" -c:v libx264 -preset slow -crf 22 -ab 128k -acodec aac -af volume=0.6 -f mp4 "Path to output file"

"-ab 48k -c:a aac" is ment to reencode the audio to 128kBit/s aac audio
"-af volume=0.6" is ment to boost the audio volume by 0.6dB
"-c:v libx264 -preset slow -crf 22" is ment to reencode the video with x264, preset slow and crf 22

I gather -map 0:v excludes the audio from the first input file, since we need to discard that un-normalized audio ?

yes, with -map 0:v you are only using the video from source 0 and with -map 1:a you load the audio from source 1, then you process the selected inputs. this is only needed in the example since audio and video come from different sources.

if anyone knows how to compress the audio range at the same time, some hints would be handy

Assuming your input audio stream is supported through ffmpeg drc routines you can use -drc-scale to influence the drc values. ffmpeg has no audio filter to apply drc during processing other than during the decoding stage.

other than that see https://forum.videohelp.com/threads/349883-Remixing-DTS-5-1-to-good-AC3-5-1-track?p=220...=1#post2202882 about dynamic range compression through sox.


----

As a side note:
- I would not encourage to use ffmpegs aac encoders (they are all not really good).
- I would prefer soxs audio processing over ffmpegs on any day
- I would use a GUI to do all that,... (e.g. Hybrid or other tools which support drc&co)

Cu Selur

[hydra3333]

Thanks for your help.

I hoped to automate the normalization, so ffmpeg "-af volumedetect" and manual adjustment probably isn't going to work for me by the looks.

If ffmpeg's aac isn't great, what would you recommend ?

I settled on this windows script to convert PAL TV captures (interlaced) for playback on Samsung Tab2 10.1 tablet):

@echo on
REM assumes input is interlaced, output is interlaced !!!

rem http://ffmpeg.org/trac/ffmpeg/wiki/x264EncodingGuide
rem "%ffmpegEXE%" -h full > z.txt
rem "%ffmpegEXE%" -formats >> z.txt
rem "%ffmpegEXE%" -codecs >> z.txt

set fINPUT=%~1%
set fOUTPUT=%~dpn1%.mp4
set fOUTPUTunnormalized=%~dpn1%.unnormalized.wav
set fOUTPUTnormalized=%~dpn1%.normalized.wav
set audiofreq=44100
set audiobitrate=192k
set crf=20
set FieldFirst=1
REM TFF

REM --------------------------------------------
DEL "%fOUTPUTunnormalized%"
DEL "%fOUTPUTnormalized%"
REM --------------------------------------------
REM "%ffmpegEXE%" -i "%fINPUT%" -c:a pcm_s32le -ac 2 -ab "%audiobitrate%" -ar %audiofreq% -f wav -y "%fOUTPUTunnormalized%"
"%ffmpegEXE%" -i "%fINPUT%" -c:a pcm_s32le -ac 2 -f wav -y "%fOUTPUTunnormalized%"
REM --------------------------------------------
REM --------------------------------------------
REM careful: dynamic range compression (compand) can make hiss more audible
REM-----
REM COMPAND values from https://forum.videohelp.com/threads/349883-Remixing-DTS-5-1-to-good-AC3-5-1-track?p=2202882&viewfull=1#post2202882
REM Film Standard
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.1,0.3 -90,-90,-70,-64,-43,-37,-31,-31,-21,-21,0,-20 0 0 0.1 gain -n -3
REM Film Light
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.1,0.3 -90,-90,-70,-64,-53,-47,-41,-41,-21,-21,0,-20 0 0 0.1 gain -n -3
REM Music Standard
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.1,0.3 -90,-90,-70,-58,-55,-43,-31,-31,-21,-21,0,-20 0 0 0.1 gain -n -3
REM Music Light
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.1,0.3 -90,-90,-70,-58,-65,-53,-41,-41,-21,-21,0,-11 0 0 0.1 gain -n -3
REM Speech
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.1,0.3 -90,-90,-70,-55,-50,-35,-31,-31,-21,-21,0,-20 0 0 0.1 gain -n -3
REM 54db Range / 5db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-54 -5 -90 0.2 gain -n -3
REM 50db Range / 6db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-50 -6 -90 0.2 gain -n -3
REM 46db Range / 7db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-46 -7 -90 0.2 gain -n -3
REM 42db Range / 8db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-42 -8 -90 0.2 gain -n -3
REM 38db Range / 9db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-38 -9 -90 0.2 gain -n -3
REM 34db Range / 10db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-34 -10 -90 0.2 gain -n -3
REM 30db Range / 11db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-30 -11 -90 0.2 gain -n -3
REM 66db Range / 4db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-66 -4 -90 0.2 gain -n -3
REM 70db Range / 4db Limit
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-70 -4 -90 0.2 gain -n -3`
REM-----
REM https://forum.videohelp.com/threads/349883-Remixing-DTS-5-1-to-good-AC3-5-1-track?p=2209539&viewfull=1#post2209539
REM Film Standard closer to Dolby
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.1,0.3 -90,-84,-43,-37,-31,-31,-26,-26,-16,-21,0,-20.25 0 0 0.1 gain -n -3
REM-----
REM lightest to heaviest dynamic range compression 
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-70,-60,-48 -5 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-70,-60,-44 -6 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-70,-60,-40 -6 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-70,-60,-36 -7 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-70,-60,-32 -8 -90 0.2 gain -n -3
REM lightest to heaviest dynamic range compression 
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-70 -4 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-66 -4 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-54 -5 -90 0.2 gain -n -3
"%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-50 -6 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-46 -7 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-42 -8 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-38 -9 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-34 -10 -90 0.2 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 compand 0.3,1 5:-80,-79,-30 -11 -90 0.2 gain -n -3
REM normalize without DRC
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 gain -n 0
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 gain -n -1
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 gain -n -3
REM "%soxEXE%" "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%" --clobber --show-progress gain -h rate 44100 gain -n -6
REM COPY "%fOUTPUTunnormalized%"     "%fOUTPUTnormalized%

REM -------------------------------------------- interlaced
"%ffmpegEXE%" -i "%fINPUT%" -i "%fOUTPUTnormalized%" -map 0:v -map 1:a -c:v h264 -preset fast -flags +ildct+ilme -ildctcmp satd -top %FieldFirst% -profile:v high -level 3.2 -crf %crf% -coder 0 -c:a libvo_aacenc -ab %audiobitrate% -y "%fOUTPUT%"
REM --------------------------------------------
DEL "%fOUTPUTunnormalized%"
DEL "%fOUTPUTnormalized%"
REM --------------------------------------------
pause
exit

edit: updated the script with the values in your link https://forum.videohelp.com/threads/349883-Remixing-DTS-5-1-to-good-AC3-5-1-track?p=220...=1#post2202882 )

Cheers

[Selur]

If ffmpeg's aac isn't great, what would you recommend ?

On Windows: QAAC (= Interface to Quicktimes AAC Encoder), FHG (= interface to Frauenhofers Encoder) or NeroAacEnc (= Neros AAC Encoder).

[hydra3333]

OK Thanks. I have NeroAacEnc (= Neros AAC Encoder) so I must look at its CLI options then.

[feelart]

Extraction seems better.

Indeed, some audio codec allow lossless normalisation (for example mp3 and use MP3Gain).

I doubt ffmpeg 1.1 would do a lossless normalisation.

[hydra3333]

Pointer to continuation in a similar thread:
https://forum.videohelp.com/threads/355502-FFMPEG-Audio-Volume-Command?p=2285229&viewfu...=1#post2285229

ffmpeg ebur128 is available with an "integrated Loudness" number.
I'm not really sure whether to use it or the max loudness number from "-af volumedetect"