Audacity Forum

Since the big lockdown, a local community radio station can't allow their volunteer DJs to come in and do their shows live. This is particularly problematic for the weekly "Vinyl Revival" show. The host would normally put together a playlist and the accompanying vinyl, bring it in to the station, and play it live.

Well now nobody can come in to do their show. I'm trying to help them keep the vinyl show going by ripping the host's playlist to MP3 files that can then be "assembled" into a show by the station. Doing this with vinyl is very problematic, especially from a time standpoint. It takes me 6-8 hours to do 2 hours of usable tracks. Clean the record, check the levels (every record is different), record the track, clean up the intros and outros, fix the worst of the "clicks", normalize, populate some basic metadata, export a master WAV file, and then export to the highest quality MP3 possible. If I'm being really conscientious, I play the entire 2 hours back in real time to make sure I haven't messed anything up that will go out over the air.

I hand off the finished MP3 files to the station who then "assembles" the show. This "assembly" process is a black box of which I have no detailed information. I suspect they're re-encoding everything and re-normalizing. It's not my purview to question their workflow or tell them to do it a different way.

The only guideline I got from the station is "Normalize to -6dB". I don't think this is a hard limit. It's just what their audio chain is set up for. Normalizing to -6dB seems a little odd to me but that's what they do and I'm trying to comply.

I've been diddling with settings etc trying to get it to sound as good as possible on-air. The latest iteration actually sounded pretty good, if not great. Of course I only have one chance a week to hear the actual results.

I've finally figured out that blindly normalizing to -6dB is just not optimal. If the peak level in the audio happens to be a vinyl "click", the entire track seems to be normalized to that peak. Not what i want.

That lead me to ReplayGain. Playing around with it I can see I'm obviously getting different results, and probably overall much better than straight normalizing. I'm recording each track with overall peak levels barely exceeding -6dB (as instructed). Then I apply ReplayGain normalization at 0dB. This seems to be working very well. Stray "clicks" are not fooling RG to anywhere near the extent that Normalizing does.

I welcome any comments or suggestions though. This is a great tool.

-dave

The new release of Audacity (Audacity 2.4.1) has a new "Loudness Normalization" effect: https://manual.audacityteam.org/man/lou ... ation.html
The default settings: "perceived loudness" -23 LUFS
is the "EBU R128" broadcast standard.

Perhaps worth having a chat with your contact at the radio station and suggest using that. As a safeguard on the first occasion, you could perhaps make them two copies - one that is normalized to -6 dB, and the other that is "loudness normalized" to -23 LUFS.

Thanks steve, appreciate the help and suggestions.

Thank you Steve for Loudness Normalization. I requested it about 2 years ago and was delivered in 2.4.1
I wish, however, that it would also allow for selection of "LRA", and "TP" in addition to "I" as defined in Loudnorm algorithm.

siamak wrote: ↑

Tue May 26, 2020 11:29 am

I wish, however, that it would also allow for selection of "LRA", and "TP" in addition to "I" as defined in Loudnorm algorithm.

LRA and TP are irrelevant to loudness normalization.

Hi Steve,

Thanks for the copious information - and apologies for reviving an old thread, but I just got hit with this problem:

I run a rock radio station at Live365.com, and after months of problems with dramatic disparities in volume between my songs and the ads that Live365 inserts periodically into the playlist, I got on one of their forums and asked what I could do to get things equalized. The reply was that I should be have "replygain tags" attached (somehow) to the songs, since that's what the L365 host site uses to normalize content-vs-ad volume. I've just done a reply at that forum - there wasn't much detail on "how" - but I'm a bit stupid about tagging things to begin with and don't really know where to start. So...

Before uploading a song to my playlist, I always take it into Audacity to ensure a uniform period of silence at the beginnings and ends of songs, and to verify that I've got the amplitude set to max headroom - with maybe a little clipping (it is rock 'n' roll after all, and square wave goodness is square wave goodness, within reason.) But despite this, the apparent ad volume is between half-again and double the song volume. Which, if a listener has got the station running in a cubicle at work, is going to get the station shut off - so it's a potential listenership-killer.

It sounds like this program only applies replygain to playback - or does it actually encode the replygain "tags" that the guy at L365 was talking about, right into the song file itself? If not, is there some separate audio editing program I need to use for this? Again, I have no clue as to how tags in general work, except in the sense of helping a web page get indexed to search engines and the like (and that's pretty hazy too.)

Any info on how to do this would be vastly appreciated.

Cheers,
SL

Usually an application that supports ReplayGain can also write the tags. But, I don't know if the tags are standardized and interchangeable between applications. I use it with Winamp. Wikipedia has a list of applications that support ReplayGain.

But there are other similar options that don't rely on tags - ReplayGain.ny (from the 1st page of this thread) doesn't write tags. It "permanently" changes the loudness of the file so it works with any application.

WaveGain (for WAV files) and MP3Gain are stand-alone applications that also "permanently" changes the loudness instead of using tags. WaveGain and MP3Gain can process a batch of files. You could probably do the same thing with Audacity and Macros. Since these things are changing your files you probably should keep a back-up of the original.