Volume Questions

[kayray]

I think the best way to deal with this is to be more vigilant in the file collection process. Perhaps making it even more important to the BC (or MC for soloists) to be sure files are at a good volume.

Lately I've been downloading all files for collaborative projects before I catalog them, and taking a quick listen in iTunes. Anything that seems too quiet (or too loud -- I had one yesterday) I open up in my sound editing software and fix. It's time consuming, but make a nicer listening experience, I think.

[a.r.dobbs]

Yes <shudder> I've just now begun the process of downloading all 48 files of Mansfield Park to my PC (mp3DiirectCut doesn't work on Mac), where I'll listen in iTunes, notate relative volume, take 'em into mp3DC and adjust gain (and fix any id tags while I'm there). Whacked.

And I ought to go back and fix some too-soft (and yes, some too-loud) files in the recently catalogued Relativity and Hollow Needle.

I don't so much mind the work as the fact that it's work based on guesswork.

[kayray]

.

I don't so much mind the work as the fact that it's work based on guesswork.

I've got an unscientific system, but it's a system nonetheless :) My laptop has wimpy speakers, so I know that if I turn the volume on a file all the way up, it ought to be pleasantly audible. If it doesn't sound "right" I take it into SoundStudio. I know how the waveforms in "good" file look, so it's easy to get them to that height with my amplify volume filter. (providing there's not a lot of ugly background noise...)

[a.r.dobbs]

Kara, you're using GoldWave on a Mac, right?
Does it take in mp3s and let you work with them without decompressing and recompressing? I believe I would invest in GoldWave if it does that.

[kayray]

Kara, you're using GoldWave on a Mac, right?
Does it take in mp3s and let you work with them without decompressing and recompressing? I believe I would invest in GoldWave if it does that.

I use SoundStudio (mac only and DON'T buy the current version (3.5) until they fix the horrible EQ bug). It does let me edit mp3s, and then save as mp3. I'm not sure whether it's decompressing/recompressing behind the scenes, but it really doesn't matter. Spoken-word mp3s can stand to lose a bit of quality.

[kristin]

I use Wavepad and can open, edit and save as MP3 but I don't know either whether it is recompressing them. But it's free.

[a.r.dobbs]

Oh, I love Freeverse, those guys are so cool!! But yeah, SoundStudio does recompress.

And the Wavepad people are awesome, too -- they have free transcription software that is Extremely Handy! unless you never transcribe, of course

Apparently the only OS X application that can edit mp3 without recompression is Rogue Amoeba's Fission (according to today's google search):
http://www.rogueamoeba.com/fission/features.php
and they have a free demo (it's $32, not horrendous) to try
and it has normalize...
but no mention of gain, so...

edit: here's bad news about the demo...

We provide a free trial of Fission for you to try. In its trial state, the quality of audio from Fission will be degraded with a series of audio fades to encourage purchase. Purchasing a license key will unlock Fission and provide you with the ability to save full-quality audio.

The good news is that it's fully functional, so you really can test it out. But bad sound quality till you buy... well... <sigh>

[rent]

I was going to mess with Villette in Goldwave but it's mysteriously vanished into the Vista mess so I need to reinstall it. I guess for anyone wanting a good audio editor with all the specs and what not, I love that one. I think it's rather expensive though, I don't quite know as my ex-boyfriend paid for it Coming from a listener perspective I guess my idea on volume is much like my idea about weather, I'd rather it be too cold because I can always put more clothing on. If it's too hot you can only strip so much. Volume much the same, I'd rather all of the tracks be loud so I have to turn it down and keep it low, than miss a single word because it's to soft.

(sorry if this is a little crazy, I work nights and I haven't been to bed yet so 1:47 pm for me is like 3 in the morning)

[a.r.dobbs]

Well, folks, I've spent the whole day long working on volume issues with Mansfield Park, struggling to make mp3DirectCut do something reasonable, muddling with Audacity, and generally, I do believe, making Mansfield Park a much more pleasurable listening experience. If it ever makes it into the catalogue.

I'm an utter newbie with this software, so this is just my limited understanding from hours of trial and error...
How I used mp3DirectCut for fixing volume on mp3 files:

• * Open the mp3 file and select all (Ctrl-A)
  * Normalize the file (Ctrl-M) and click OK (it remains all selected)
  * Adjust gain: Ctrl-G, important: check box "Lock sliders" and slide gain up +6 to +12dB for soft to faint files, or down for distortingly loud files. Click OK. Listen to the file* and hit Ctrl-G again if you need to adjust the gain. If it's good, SAVE; it doesn't really apply to the file until you save.
  * Save (Ctrl-W). If you save with the same name in the same folder, it will overwrite itself by closing first and then saving. You'll have to open it again to continue working.
  * If you need to add gain to the whole file and then fade to silence at the end, you need to apply one gain change, save, open the saved file, and apply the second gain change, and save it. You can't make two different gain changes without saving and reopening between changes.
  * If you need to make a cut, save the file after the edit, and open the saved file before continuing with the gain change; otherwise, the edit point from your cut will interfere with the way the gain is applied.
  *To undo, the only method I know is to reopen your file without saving it this time: oops, how I wish I could Ctrl-z, but no; well then, I'll hit Ctrl-O to open a new file, say NO to "save changes?", and then reopen this same file.
  * To add blank space at the end: I had to select (drag your cursor) and copy (Ctrl-C) a 5- to 7-second piece of the file and paste (Ctrl-V) it at the very end. Then I had to save the file!!! I then reopened that file and selected the pasted-on portion, hit Ctrl-G for gain, selected the "Silence" button, and saved the file. Watch out! If you miss a step here, DC may make your Whole File silent! -- in which case, use the "feaux-undo" above.

That'll do for now.
Marching on here...

*When you hit play or space-bar, DC plays what's at the center of the window; the dotted line shows what's being played. Drag the blue tab under the window. If you hit play and then drag the tab, you'll "scrub" the audio (just meaning you'll hear it at your dragging speed, whatever is on the dotted line).

[Planish]

We need a program that

• * lets us work within the mp3 without re-compressing it [the data] on the new output
  * evaluates each mp3 to tell us its overall db level
  * allows us to elevate or decrease the db level without clipping

And we need a target db level to aim at.

When you use the word "compress" you should make the distinction between audio compression and data compression.

With audio compression, (and there is an Audacity effect called "Compressor") you can increase the average or perceived volume such that the peaks are not amplified to the same degree as the rest, so you don't get clipping. Any low-level background recorded noise is amplified as well.

Many commercial radio stations do this to increase the apparent volume without over-modulating the RF carrier. A station that caters to the high-fidelity fans (such as a classical music or jazz station) might use very little or no compression, so as to preserve the dynamic range.

The problem with trying to decide on a common reference level is that speech is not a continuous tone, it has dynamics. A decent audio level monitor should have two meters - one that is a Peak Program Meter which responds quickly to transients (such as loud clicks and pops) and sorta holds it at the maximum for a second or two; and another that is a VU meter ("Volume Unit") which responds differently. The VU meter doesn't hold the peaks, and it doesn't deflect as much as the PPM, so it is a better indicator of how that level would be perceived by a human ear. Some digital meters might have a bar that grows and shrinks according to VU "ballistics" but also have a little dot that floats at the PPM levels.

So, when adjusting levels, you watch the VU meter to find an "average" perceived volume, but you watch the PPM to prevent clipping by the recording or broadcasting equipment.

A simple "normalize" function can be bad, if the majority of the program is about average and there is only one or two transients. That would push everything down.

What we need is something that provides the effect of slow AGC ("Automatic Gain Control") which looks at the average level over the past several seconds worth of audio combined, with some kind of compressor/limiter that prevents clipping when loud transients above some arbitrary threshold level suddenly show up.

[a.r.dobbs]

Right you are -- I've changed my post to specify "the data."

Since we don't have a sweet, simple piece of software to tell us scientifically what volume changes to make, we need a nonscientific, a subjective way that works. As a few of us wade into the trial-and-error method, we can share our discoveries to help other folks who'd like to fix the volume fluctuations in collaborations.

Kara is going to create a few sample files of various volumes and show their waveforms. That way, folks can make a test of their setup and compare their own waveform to the sample picture and adjust their setups as needed.

For BCs or MCs tackling volume, we could use Kara's great-volume sample as the target we're adjusting the project files to. Or we can choose the most prevalent good-sounding volume level in a given project and adjust the other files to that.

Here's one way:

• * open the model file in iTunes (target volume)
  * open the file to be adjusted in mp3DC / Audacity / whatever
  * make adjustments till the two are a very similar volume (listen to one and then the other, and sample all through the adjusted file)

I recommend using headphones and setting the volume controls -- in iTunes, in the editing software, and on the computer -- to a notch or so below maximum (and always use these same volume settings). We want listeners to be able to turn the volume up in noisy environments, so we need that headroom for them!

Well...those are my thoughts right now...

[Planish]

You're right. Ultimately, it comes down to making a subjective comparison to some reference file. That's why movie soundtrack editors get paid the big bucks.

Here's one way:

• * open the model file in iTunes (target volume)
  * open the file to be adjusted in mp3DC / Audacity / whatever
  * make adjustments till the two are a very similar volume (listen to one and then the other, and sample all through the adjusted file)

Or...
Here's an idea, using Audacity for the example. (I just made this up.)

• * Open the file-to-be-adjusted in Audacity, and have it play back (through stereo headphones) on just the left channel, using the on-screen L/R balance control for that track.
  * Paste the model file into a second track in Audacity, ("Project" menu / "New Audio Track") and have it play through the right channel only, using the balance control for that track.
  * Play a selection of both tracks simultaneously. Listen to see if one reader sounds subjectively louder (or "closer" in the stereo image) than the other.
  * Adjust the playback level for the track-to-be-adjusted, using the +/- volume control for that track only, such that the two readers seem balanced with each other. As you slide the volume control, the gain in dB (eg. "+6dB" or "-18" appears. Make note of it because that indicates by how much the track-to-be-adjusted needs to be adjusted to match the model track.
  * Once the adjustment is done (using whatever combination of Compressor and/or Amplify effects you use) delete the model track, reset the playback balance and gain of the adjusted track to middle settings, and export.

If the required amplification would results in clipping, use the "Compressor..." effect beforehand to decrease the transients (pops, clicks, plosives) without increasing the average level. Be sure to de-select the "Normalize to 0 dB after compressing" option.

[a.r.dobbs]

Oh -- I should put in my Audacity how-I-increase-volume notes here too. Can't right this second. Will later.

I never use Amplify ... almost never ... in Audacity.
I have a much more complicated method.
Really, much more flexible and fun and nice in the effect, imho.
Gotta go, tho. Later.

[jimmowatt]

Oh -- I should put in my Audacity how-I-increase-volume notes here too. Can't right this second. Will later.

I never use Amplify ... almost never ... in Audacity.
I have a much more complicated method.
Really, much more flexible and fun and nice in the effect, imho.
Gotta go, tho. Later.

I'll bet you copy the track and then merge them don't you.
or
You boost frequencies in the equaliser that enrich your voice.

[a.r.dobbs]

Mmmm. Duplicate the track and ease the gain up if needed. When volume changes suddenly, I silence out the louder part on the second track. I might fade it out to silence. I might make a 3rd track and equalize it and reduce the gain so the effect is subtle. The potential complications are nearly infinite! When you output to wav or mp3, the tracks are instantly merged.