• Get involved.
    We want your input!
    Apply for Membership and join the conversations about everything related to broadcasting.

    After we receive your registration, a moderator will review it. After your registration is approved, you will be permitted to post.
    If you use a disposable or false email address, your registration will be rejected.

    After your membership is approved, please take a minute to tell us a little bit about yourself.
    https://www.radiodiscussions.com/forums/introduce-yourself.1088/

    Thanks in advance and have fun!
    RadioDiscussions Administrators

Determining Maximum Audio Level

G

GeneSavage

How do you determine what is the loudest volume that you can feed your stream?

I'm using Stereo Tool followed by Sound Solution (long story) and have always gotten complements on my sound EXCEPT they often comment it is not as loud as other internet stations.

Indeed, if I record the stream from Winamp at 100% volume, my level is down about -6db.

My output on Sound Solution is set to 100%, and the peak clipper is set at 0db. It flashes regularly to indicate it is stopping peaks beyond 0db.

I've heard that it is important to use every db of the codec so that it is encoding as efficiently as possible, but I've also heard digital clipping is NASTY.

When I shut off the clipper and crank up the output, I do eventually start getting a gritty, distorted sound, but it's more like an analog tape overload than a digital clip. I wonder why?

Pushing it as hard as I can, I can see tremendous density in the recording, but can't get it to peak out past 98.51%, or -0.13db. While I agree that's VERY close to 100%, that's with everything overloading... why doesn't it come out at 100%?

MOST importantly: how can I determine how hard I can push the encoder without creating new distortion by overloading either the output of Sound Solution or the input of Edcast (my encoder)?

The meters are so small it's hard to do precise measurement. I can tell I'm in the red, but there's no "peak" light.

Can anybody help?
 
I normalize my stream to -3. In theory, you could run it up to -1 or -.01, but there are MANY crappy soundcards with lousy DAC (digital to analog converters) that distort easily. So I usually have it pegging at -3 to leave some breathing room...-6 is a bit too low IMHO.
 
OK, so SOME overhead is a good thing; makes sense.

My biggest problem has been determining my actual, exact volume. My current process is to play it through Winamp at 100% volume and record via the "software" setting in Total Recorder. I then go back and view it in Adobe Audition. It's a long, drawn out process that means changes take 15 - 20 minutes to do effectively.

In doing some reading, I'm hearing that a look ahead limiter is the best for final level protection. I was originally thinking I HAD to have distortion-cancelled clipping, but the documents I've found online suggest look ahead limiting can get me absolute control over my peaks with zero distortion.

To that end, I'm testing out a free VST plugin right now simply called, "Look Ahead Limiter." It's single band not multiband, but I assume for final peak protection that's OK.

The biggest challenge for me is that I'm using Station Playlist Studio, which uses Winamp plug-ins... which means I'll have to use a VST plug-in to run the limiter... and in the past I've found those a bit unstable.

...Never mind that I already have Stereo Tool and Sound Solution running! lol... I'm playing a game of Jenga over here with my plug-ins.

Money would make all of this easier as I could buy a pro processor and be done with it... but the budget (according to my wife) is zero dollars and zero cents... so the search continues. :)

Thanks for your input! How do you determine you're hitting -3db peaks?
 
Skimp on something, somewhere and save up 30 dollars for Breakaway personal audio processor software.

It's not quite as stunning as the x10 more expensive broadcast version, but it's very, very smart
and does not do "funny" things like so many plug-ins.

Leave 2-3 db headroom and be amazed. ;D
 
NightAire said:
How do you determine what is the loudest volume that you can feed your stream?

I'm using Stereo Tool followed by Sound Solution (long story) and have always gotten complements on my sound EXCEPT they often comment it is not as loud as other internet stations.

Indeed, if I record the stream from Winamp at 100% volume, my level is down about -6db.
Click to expand...

What a listener comments that they hear may have no relation to what the meters tell you. Loudness can be a personal perception. Maybe the listener has their equalization of certain frequencies cranked up and you have less energy in that frequency range that some other streamer. You can go crazy "chasing your own tail" if you can't control the method of determining loudness.

Get your listeners to give you the identity of streams that they perceive to be louder. Record the other stream(s) just as you have recorded your own stream and see how they compare.... both on the visual.... and in your ears. Then ask youself: Do I agree the other guy is louder?

When you talk about examining a recording from your own stream, and you describe your levels as being minus this or minus that (as in -6db) are you describing the PEAK levels? Measure the RMS value of your signal versus the supposedly louder signal and see if they have a compression/limiting system that makes their signal more "dense".

I understand the mental process in the broadcasting world where a program directors pictures someone in their car punching the buttons to switch from station to station to station. Perceived loudness may be a bit more important there. But in the world of digital audio in a streaming world, is a 6 db difference in signal really that critical? In the broadcasting world, the atmospheric noise level remains constant as the audio signal varies from station to station. But in the world of digital audio streaming, what is the penalty for being a few db lower. If the listener likes the content, they turn the gain up just a bit and everyone becomes equal..... or do they?

I do a lot of editing of recordings of live events: lectures, recitals, wedding, funerals, sermons. They are usually recorded via a CD recorder. When I rip them and open them up in the editor, now and they my heart beats a bit erratically when I see the peaks are -20 or maybe -25. But in the digital world that doesn't creat a significant problem because the room noise and system noise are way down there also. Raise the gain on everything, take out room noise and system noise as able, and the final result is identical to a recording that I receive with peaks at -1 or -3 or -6. (I take the room noise and system noise out of them also.) How is streaming audio that much different than that?
 
Goat, I'm right there with you: if it was just a density issue, I wouldn't sweat it. What I'm measuring is peak output sent... and some of those who have critiqued the levels are people I respect within the industry.

The theory was, if the encoder didn't see something near 100% volume, I was "wasting" some of the codec's ability to determine what to keep and what to throw away. (That's one of the problems of over-compressing and over-limiting, too: if EVERYTHING is 100%, how does the codec decide what to throw away?) I was "wasting bits."

What I couldn't figure out is how I could have the output of Sound Solution at 100%, have the clipper lights occasionally flashing at 0db, and NOT see anything close to that kind of output on the other end. I still don't understand it, but putting a look ahead limiter as the last thing in the chain seems to have eliminated the output problems. I'm using less than a db of limiting, but now seeing as much output as I can stand; in fact, I've actually backed the output of the limiter down to -1 db! I little bit of limiting goes a long, long way.

BTW, in case you've read the above and gotten worried: I shut OFF the clippers on Sound Solution. :)

I think if I was streaming linear audio I wouldn't worry about it; since I'm streaming compressed audio, I'm a little more concerned. (Perhaps unnecessarily...)
 
I'm going on the assumption you are talking about BlackLightRadio... that appears as a footnote in your post. I went out, I listened, I captured some audio and looked at it in Adobe Audition.

Sounds good.

I think I see what is bugging you. A lot of peaks at about -6, quite a few at about -3 but a spike here and there at -1.

I measured the "density" of the newscast (I wanted some spoken word) so I could compare it to the density of some spoken-word recordings I have processed for podcasting. Yeah... I guess with the right combination you can squeeze out about two more db, and maybe three if you have super-human powers.... and a fat wallet. But I do it on a recording that I can process multiple times if necessary. The ability to do it "on-the-fly"... on a live stream.... beyond my pay grade!

I spend a lot of time learning how to squeeze out just a bit more but I come away asking myself... WHY! Which is the same question I pose to you. And the answer comes back: "Because I can, I will!!!!"
 
Cowboy (sorry I called you "goat" earlier; I wasn't thinking then, & was coming back to change it!),

lol... I agree. We battle to get every drop of sound sweetness out of something, yet I've never had a listener say "I'd listen to that station, but they're just a db or two too low." ;D

Yes, BlackLight Radio is the stream I'm massaging; thanks for the kind words! It's a lot of fun. I'm hoping someday it will be a source of income... but for now at least it's fun. ;)

I don't think I want to try to squeeze anything more out of the signal, at least not without a pro (paid) processor. I can decide if it's psychological or actual, but it feels like I can "feel" the audio hitting the limiter... or I'm hearing the codec "fill up" with audio... not sure which... and it make the hair on the back of my neck stand up just a little (not in a good way). I'm going to a give a week or two so I can "forget" it's in the chain and see if it still bugs me, or if it's because I know it's there that it bugs me.

Before I had the limiter on there, I would NEVER get above -6 db on peaks, even after 15 minutes. I still don't understand why Sound Solution wouldn't give 100% output, but I've "fixed" it now.

Now I just have to decide if it's TOO much.

You may have experienced this with some of your recordings: given the right combination of audio, I've seen an mp3 codec clip if you had a peak of -1.75 db. It was likely overkill, but when I was using lossy files for my stream (heaven help me!), I would maximize all of the files to -1.75 db just to make sure NOTHING clipped.

I'm thinking out loud here: I'm wondering if setting the output of the clipper to -1.75 db might be a good, safe maximum setting...

I know, I know: beyond your pay grade! :D I appreciate your responses and feedback. Every ear helps! (It also helps that I've learned to value your opinion as I've seen your other posts across these boards; I always feel like I learn something from you, whether I agree or not.)
 
NightAire said:
Cowboy (sorry I called you "goat" earlier; I wasn't thinking then, & was coming back to change it!),
Click to expand...

"Goat " is a totally acceptable greeting. :)


NightAire said:
You may have experienced this with some of your recordings: given the right combination of audio, I've seen an mp3 codec clip if you had a peak of -1.75 db. It was likely overkill, but when I was using lossy files for my stream (heaven help me!), I would maximize all of the files to -1.75 db just to make sure NOTHING clipped.
Click to expand...

You have just reminded me that I have NOT done any testing where I push audio up to maximum value, save as .MP3, and then bring the .MP3 back into the editor for really close examination. I listen to my MP3s and my old (in GOAT YEARS) ears tell me the sound is acceptable, but if I decode the audio and expand the wave form on screen, will I see some really ugly things? Details of that little test coming soon.
 
For my stations internet stream I never let it exceed -6db.
It usually runs about -7db with peaks at -6db.
For me anything higher is in the "yellow" on the encoder.
Above -6db I run the risk of clipping and distorting.

I can crank all my inputs as high as I want and it will never go above -6db.
I can also turn my inputs as low as possible and it will drive the audio back to -6db. (Thanks to the AGC/Limiter)
Too much input can cause distortion for me.
I guess some are different, but I honestly can't tell the difference in audio loudness between my stream and others.
The only ones that appear louder are the Clear Channel streams.

That being said, Internet radio isn't FM radio, we don't need to be loud!
 
I use Breakaway Broadcast and my encoder (edcast) takes the feed directly from Breakway. I set the attenuation in edcast to -3 to bring it down since Breakaway outputs it at 0 db.

Is -3 too loud? Maybe. But I like a BIG processed loud sound that jumps out - and for the format (jammin oldies), it fits. http://jammin.servemp3.com is the stream URL (128k AAC) if you want to take a listen. I love how it sounds in a car - someone told me it reminded them of NYC FM's in the 80's.
 
I come away from this discussion with this observation: Software and hardware in the price range we are talking about is not likely to be that accurate. When you say you want -3 or -6 out, do you know that is actually what you are getting?

To me the most trusted piece of software I have when it comes to measuring things like this is Adobe Audition, and I have learned there are some "measurements" in that software that I can trust, and some are just estimations.
 
To quote former Prez Ronald Reagan, "trust, but verify." And I did - the attenuation knocks it down -3 db like it should. If I had the money, I'd buy a AirAura and do everything on a hardware box...but since 14k is roughly 60% of my yearly income, Breakaway Broadcast will have to make do for now. I know the levels going out of NexGen & into Breakaway are spot on because I'm using a Digigram soundcard with XLR in's and outs and a metering program to make sure everything is aligned right.

That's one of the more important things - having a decent pro grade soundcard. Using the built in 1/8th line in is just asking for trouble...
 
Breakaway Broadcast is no slouch! There's a local FM here using it on-air, and it sounds GREAT. It is extremely accurate... accurate enough that you can drive a transmitter with it if you have a pro card capable of 192 Khz sampling rate.

My software is all freebee stuff... and I do it all in-system... the audio never leaves the digital domain (or the box, for that matter). I run Stereo Tool, Sound Solution 1.31, and Lookahead Limiter as DSPs, and feed it to the Oddcast encoder.

Technically, I could stream without a sound card at all, but I like to monitor straight off the box....

Liberty, two questions: #1, what are you using for gain control, and #2, you get distortion over -6db output? That doesn't sound right...
 
Gain control is handled by the simple little plugin EnergizeII, which also limits. I'm not sure if it actually distorts above -6db, but thats where the SAM Encoder plugin turns yellow, so I leave it down there. That being said, I could probably turn it up a hair.
 
LibertyNT said:
That being said, Internet radio isn't FM radio, we don't need to be loud!
Click to expand...

Yes, loud isn't the issue, it is the clarity of the sound. Listeners can push up the sound a bit if they really want.

If you are using lossy formats for songs, you are already losing some audio quality, but that isn't what is being talked about here. I have mp3, m4a, wma, FLAC and WAV files in my regular rotation, but notice a big difference between the WAV files (FLAC too) as compared to an mp3 file. Most of what I have in the past few years is in WAV format. Need a bigger drive to accommodate, but worth it in the long run, as it pertains to sound quality.

Some sound processors do a good job with clipping above acceptable levels. Breakaway is one, and Stereo Tool is another. Sound Solutions doesn't do as good a job in my opinion.

I think if you have a decent sound, the listeners will be more likely to comment on your content than the overall sound. Having a decent sound output is important, but content is so much more so. I've never had a listener comment on the overall sound, but many on the content. Just a thought.
 
Liberty, that yellow should be just warning you you're getting close to the peak... not that you're actually peaking.

As has been said here, it's not critical that you punch it all the way up, but I find cheap, weak sound cards like a stronger signal (where listeners have to turn up the volume less). If they have to turn up the volume TOO much, the can start hearing internal noise from the cheap sound card, and I've also noticed it can sound "thin."

You might try a db or two more, at least...
 
Status
This thread has been closed due to inactivity. You can create a new thread to discuss this topic.


Back
Top Bottom