Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: workflow: volume problems in merging audio books & radio plays with DSP (Read 920 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

workflow: volume problems in merging audio books & radio plays with DSP

Hi,

i want to merge my lossless .flac radio plays which i have "mostly" with multiple tracks in hopefully constant intra-album Volume in a batch process.

my goal ist to transcode all for a walkman. I just need them for a child and in 1-2 Years they will all be unnescessary.
I don't want to collect them.

I tried foobar:

1. Step: normalization with ReplayGain albumbased to somewhere between 0,85-0,92

considerations and problems: I don't know or trust the mastering, if the volume from track to track is constant. I have bad experience (I find upscaled titlesongs, even ripped from CD overcompressed alternate with wide dynamic tracks, etc.)

2. Step: DSP: i want to add "Vocal Exciter" around 0.15 because it enhances for our ears and hardware the llistening pleasure.
consideration: The vocal exciter generates more dynamic range an the peaks could generate clipping and distortion

4. Step: DSP Advanced Limiter or EBU or ReplayGain Album before transcoding
There is a big difference in the transcoding output in size if the track volume before transcoding is 0,5 or 0,9 from full-scale.

The player in car or mobile have no replay-gain support or should not be dependend on it. If sometimes played on a PC or smartphone, 1by1 will be the player (with carefull normalization).

5. transcode to e.g. tvbr qaac q64

6. Step: everything merged as Multitrack with chapters in one file

How can I make it happend, that the Volume of the end-track is in a safe region not to loud - nor to quiet.

So it sounds very simple just:
- validate/trust the original files peak-buffer for the process.
- Adding Vocal Exiter
- normalizing before transcoding
- transcoding for simple analog hearing without distortion or too quiet or loosing too much Dynamic Range, if a compressor will be suggested

Problem: checking with RG after transcoding gives me very low ReplayGain values.

I tried a lot combinations, e.g. a 40 Track Radio Play has uncompressed ReplayGain Values from 0,96-0,99 TP, RMS -15,5, Peak: -0,07
the process was:
ReplayGain -2db apply albumbased/prevent clipping, DSP: Vocal Exciter, Advanced Limiter
the output was: ReplayGain Values from 0,47-0,71 TP, RMS -21,7, Peak -3,0

How is the processing chain in foobar?

Are DSP EBU/Limiter for a whole album or just track based?
Comes Replaygain after DSP or before?

How can I make it happen, that the Volume of the initial, processed and end-files are in a safe region neither to loud nor to quiet?

Re: workflow: volume problems in merging audio books & radio plays with DSP

Reply #1 –
Don't tell anybody here but I'm not a foobar2000 user...  ;)

Quote
1. Step: normalization with ReplayGain albumbased to somewhere between 0,85-0,92

...The player in car or mobile have no replay-gain support or should not be dependend on it. If sometimes played on a PC or smartphone, 1by1 will be the player (with carefull normalization).
You can use an audio editor such as Audacity (free) or GoldWave ($50 USD) to adjust/match LUFS loudness.   That way you "permanently" change the loudness of the file and your player doesn't have to support ReplayGain or do anything special.  (LUFS uses a digital dB reference so your dB levels are in the ballpark of -20dB instead of +89dB.)   

Just like ReplayGain or any other loudness-matching there is a compromise involving how loud you can go and still match volumes without clipping. 

Generally, the best procedure would be to normalize (maximize) everything first, then choose the quietest track as your reference and adjust the other tracks down as necessary to match.   (That way you avoid clipping.)

O, you can use a similar method by-ear - Normalize/maximize everything.  Then choose the quietest track as your reference and adjust the others down to match. 

Quote
albumbased... 
Are you sure?    Album based means one adjustment to the whole album so loud songs remain relative loud and quiet songs remain relatively quiet.

Quote
4. Step: DSP Advanced Limiter or EBU or ReplayGain Album before transcoding
There is a big difference in the transcoding output in size if the track volume before transcoding is 0,5 or 0,9 from full-scale.
A limiter should prevent hard-clipping (depending on the settings) but limiting is a kind of dynamic compression and it can sound like distortion if over-done.   Typically, limiting is followed by "make-up gain" or normalization to bring-up the loudness.   But then ReplayGain might bring the loudness down to hit the target setting.

Quote
6. Step: everything merged as Multitrack with chapters in one file
That's just not standard for "computer files".    If you buy an MP3 album from Amazon (or an AAC album from iTunes) you get a separate file for each song.  The files are "tagged" with a track number so the player can the files in the correct sequence.    Or, you can make a playlist if your player supports that.

Quote
How can I make it happend, that the Volume of the end-track is in a safe region not to loud - nor to quiet.
What?    ReplayGain (track gain) or volume-match before "merging".

Quote
Problem: checking with RG after transcoding gives me very low ReplayGain values.
Maybe from "excess" limiting?  Actually, limiting should allow you to go louder overall but you might have to boost and volume-match after limiting. 
 
ReplayGain (or volume matching) should match your target (possibly limited by clipping) if you do it last.

Re: workflow: volume problems in merging audio books & radio plays with DSP

Reply #2 –
Thank you for your advice. Thera are som isunderstandings and confusions. I have found a few similar threads and will come back after understanding and further tests.