I’ve used audacity for years and never had any issues getting it to do what I needed, but right now I am working on some music which I’m producing from sample libraries with the music is being exported to wave files, and no matter what I do to them in Audacity, the levels are just too low. Whenever I go to play the recordings through any audio system (my car, a stereo, etc.), I need to turn the volume up very high to get a decent volume - this includes both the soft parts as well as the loud parts of the music. Obviously the loud parts are louder, but the volume knob still needs to be turned up high to get the volume I should.
I’ve tried using Audacity’s amplifier to got up to 0.00 dB, but that is only marginally better. I’ve tried running it through the compressor with a variety of settings, and while that is somewhat better, it’s still not great - and has the cost of really squashing the highs and lows. At this point, I’m very confused because I’ve opened up several film score tracks I’ve imported from CDs - John Williams, Jerry Goldsmith type of stuff - and they look very comparable to the file I’m trying to fix up, but sound much louder. Comparing the file I’m working on and some of these imported tracks, both hover pretty close to 0.0 dB on the decibel scale, and the imported tracks are much lower on the linear/vertical scale.
Have you tried Chris’s [dynamic range] compressor: it’s better than the one which is shipped with Audacity …
I’ve written a program that makes it easier to listen to classical music, or other music that has a wide range of volumes, at low volumes or in high noise conditions (such as in your > car> ) so that you can still hear the soft parts. I’ve also written a plugin version, designed to be used with the free audio editor Audacity.
I will be trying that when I get home, but I don’t know how much help I expect it to be. Like I said, as it is even the loud parts just aren’t sounding really even at an average volume.
By the way, when I do test this in my car I will do so at rest with the engine turned off. I always check each recording at least once in the car just because it has a better audio system than anything in my house.
It’s a plugin. He wrote it so he could listen to opera with its large volume variations in the car. One of the side effects of the compressor is the show gets much louder with little or no apparent damage.
Live music captured from actual, physical musical instruments usually have dramatically large volume ranges, particularly if the performer is being “expressive.” It doesn’t take very long for people to discover those don’t sound anything like carefully and expensively produced commercial music.
I’m going to give Chris’ compressor a try. Any thoughts on why the professional recordings I’ve checked out sound louder, but have waveforms that look smaller than what I’m working on?
Human hearing is not equally sensitive to all frequencies … Equal-loudness contour - Wikipedia
e.g. A 1KHz sine wave at 0db will be loud and clear, a however 20KHz sine wave at the same amplitude will be inaudible (except by dogs ).
Loudspeakers also have a bias towards certain ranges of frequencies: even if the amplitude of the signal fed to the loudspeaker is constant, certain frequencies will be reproduced louder, (due to resonance).
That’s what the producer is doing between the time the artist records the performance and the CD goes out the door. There’s usually a several week gap in there where the producer and engineers “get it right” with all those tools in the studio and/or mixing programs.
I did an April Fool thing a while ago (in April) where I introduced the “Professional Audio Filter.” In it, I distilled all the things that people always posted they wanted. You put your show in there, push the button and it comes out sounding like it was mixed at Abbey Road Studios. The gag was a huge hit on the video forums, too. They have the same problem made much worse by the presence of pictures. It looks stunning. The sound should automatically come out good, too, right?
The truth is it’s a career move to know how to mix like that. The best we can do is collect the tools that we guess would work based on knowing what the finished show sounds like. You then become a cook trying to figure out how much salt goes in and when. Chris’s claim to fame is a compressor tool that makes the show much louder and evens out the variations between loud and soft passages, and in addition, doesn’t seem to be doing anything when you listen.
Yeah, I hear you… I’m really not concerned with blowing anybody’s speakers out, I just want people to be able to hear the music without having to turn the stereo up 3/4 of the way. I’m trying to get around the problem of amplifying it to 0 dB not being loud enough to hear, but amplifying it past that obviously causing distortion.
I’m basically agreeing with Koz’s last post…
“Dynamics compression” is partly about using the right tools, but it is definitely an “art” to get just right.
Chris’s dynamic compressor is very highly rated on this forum (I’ve recommended it many times) because it does a particular type of dynamic compression really well. if you read his blog page he explains that the original motivation for the effect was to even out the loud/quiet passages in (classical) music. He has succeeded. His plug-in also provides a very high quality “radio type” compression that is well suited to raising the overall level and evening out volume changes. However there are different types of dynamic compression - different “flavours” if you like. If Chris’s compressor is “vanilla”, there is also “strawberry”, “chocolate”, “beef” and many others.
If Chris’s compressor does not provide the “flavour” that suits your production, there is also the Audacity built-in compressor effect, the “Dyson compressor”, Steve Harris’s “SC4 compressor”, the Steve Harris “Fast Lookahead Limiter”, and many others.
For a very different flavour to Chris’s compressor, I’d recommend trying the “Fast Lookahead Limiter” (available in the 90 plug-in LADSPA pack http://audacityteam.org/download/plugins ). Chris’s compressor makes fairly gradual changes to the “gain” (amplification), whereas the fast lookahead limiter operates very fast, but only on the peaks. The two effects can be used in combination to good effect - a modest amount of compression with Chris’s compressor followed by a bit of “peak limiting”, and then amplified close to 0 dB can increase the overall volume considerably.
And please note that no matter what you do, you are remixing the character of the song and it’s not going to sound the same. Louder and desirable, yes, but not the same. It’s not like turning the volume control up. When I simulate the sound of the local NPR FM station, I use settings the the elves generated where I increase the compression from 0.5 to 0.77 and leave all the rest of the setting alone.
Chris rejiggered his control panels a bit ago and I haven’t used it since then.
You are using a fully produced sample of a commercial mix as an example. What you really need for a good comparison is to hear the commercial mix before they messed with it.
Well this is what I am trying to achieve - is to get something which is as faithful to the original sound as possible but which can actually be heard. Chris’ compressor just isn’t doing it. I’ve tried it with the default settings, I’ve tried it with the .77 ratio, and I’ve tried it with various other settings. It messes up the frequency response in all of these cases and completely changes in some cases what instruments in the mix I can even hear.
You are using a fully produced sample of a commercial mix as an example. What you really need for a good comparison is to hear the commercial mix before they messed with it.
Koz
I suppose this is true, but all I am trying to see from the example is that how much headroom they left, what amplitude looks like, etc., relative to what I’m working on. I’d expect recordings of similar instruments both of which are at -3 dB to sound just as loud as one another. When I listen to some violins in one mix with the amplitude hovering around -3 dB and then some violins in another mix with an amplitude also hovering around -3dB, but one sounds twice as loud as the other, I wonder, gee, what makes this one sound louder, and how can I achieve that?
Did we ever find out the particulars of your system? I don’t recall it. Windows? Which one? What are you using to burn the CD? Which program and what are the settings? I recall a setting in iTunes where you can set automatic level compensation where it tries to set the volume for you.
Over on Windows, you have another interesting problem. Modern Windows machines have audio conference settings which will try to set sound levels for you – whether you want them to or not.
We are now in the position of trying to describe how blue something is, you probably should post actual music.
Select about five seconds of those violins in your example, Tracks > Stereo Track to Mono, and export as FLAC with default settings. Attach both files, one good, one bad to your next post. See the bottom of the text entry page: “Upload Attachment.”
I’ll get a file up once I am back with the computer that has the recordings.
As far as the system, it’s Windows 7 x64 running on a quad core 2.8 GHz AMD processor, 10GB RAM, 7200 rpm HD. It’s only using the stock realtek soundcard right now (which I will be upgrading after Christmas), but it sounds fine playing back through the system. It’s dealing with trying to export that audio so I can burn it to a CD, put an MP3 on the internet, etc. that is a problem. I’m working with wave files right now, of course. Once I have it right I’ll make an MP3 out of it for internet distribution.
Ummmm. I think I got lost there. It’s only during Export that the problems manifest? How do you know that the waveforms are the same between similar performances of loud and soft songs?
I’ll get a file up once I am back with the computer
I’m sorry, I’m not sure whether I understand your question.
Let me just explain in greater detail what I am doing, and if that doesn’t give you the information you need, ask me whatever else you need.
I am creating audio files with orchestral sample libraries (mainly East West’s Symphonic Orchestra). This is done in a notation program. When I play this back in the program, it sounds fine. I then have the Notation program then bounce the audio to a wave file, which I open up in Audacity to prepare and, for the internet version, export as an MP3. If I just take that wave file and listen to it on some audio system (stereo, car, etc.), it sounds too soft, and so I have been trying to amplify it, compress it, or do whatever processing is required to boost the loudness to an acceptable level (I’m not looking for modern “radio-ready” levels here, just ones I can hear with reasonable clarity without having to turn the stereo up 3/4 of the way or more).
Now, as I began to attempt this and found that I could never quite get the loudness that I felt I should, I wanted to look at something similar which I thought had appropriate levels to see what they had done to it: was it so compressed that the entire waveform looked like a solid bar, as I’ve often found with pop music? What was the peak level that that was used in that recording? Thus, I took some files I had imported from the Apollo 13 soundtrack, from one of the Star Wars soundtracks, etc. What I found was that they looked relatively uncompressed (though I am sure something was done to them), and that the peak level in those tracks was about -4 to - 6 dB. I also was surprised that if I took a section from my file with peak level of 0, it would sound at best as loud and at worst less loud than the imported tracks.
I understand the amount of processing that has probably gone into these tracks, I just don’t understand why exactly similar instrument sounds/frequency ranges at greater levels in one file don’t sound as loud as those at lesser levels in another file.
).