avatar

AudioArmada

New Forum Friend

Posts: 4 Member Since:19/08/2014

#22 [url]

Aug 20 14 9:46 AM

tom eaton wrote:
Also a heads up that Sample manager saves files at their original bit depth... so if you plan to SRC a 16 bit file, convert it to 24 or 32bit first.

For what it's worth, I'm fairly confident that the iZotope SRC will increase the wordlength to at least 32-bit (if not 64-bit) before it actually executes the SRC.

-Corey

Quote    Reply   
avatar

tom eaton

Platinum Blonde

Posts: 1,334 Member Since:25/01/2011

#23 [url]

Aug 20 14 9:55 AM

"Fairly confident" is not what I want to hear from the manufacturer! Ha, ha...

The point is that if you do any process to a 16 bit file sample manager renders it back to the hard drive as a 16 bit file, even if the process resulted in data that was greater than 16 bit.

Quote    Reply   
avatar

AudioArmada

New Forum Friend

Posts: 4 Member Since:19/08/2014

#24 [url]

Aug 20 14 10:16 AM

Well, we license the SRC, so we can't know everything about their IP. Double-checked with our engineer, though, and internal processing occurs at 32-bit, but yes, the result will be at the bit-depth of the source. Always a good idea to be explicit about your bits :)

-Corey

Quote    Reply   
avatar

rbreen

Silverado

Posts: 91 Member Since:31/03/2013

#25 [url]

Aug 20 14 10:37 AM

AudioArmada wrote:
...That's another reason to use AAC, FLAC, ALAC.

-Corey

So Corey - are you saying the AAC encoder is capable of accepting 96k files natively?  In Apple's case at least, they insist on SRC before conversion, but are there some that don't?

best,
rich

Quote    Reply   
avatar

rbreen

Silverado

Posts: 91 Member Since:31/03/2013

#27 [url]

Aug 20 14 11:28 AM

AudioArmada wrote:
Actually, I was mistaken there. AAC (standard) only supports a max of 48k, as well. 

-Corey

OK great - that's what I thought - just wanted to make sure I wasn't missing something...

best,
rich

Quote    Reply   
avatar

aleatoric

Gold Finger

Posts: 258 Member Since:01/02/2012

#28 [url]

Aug 20 14 12:01 PM

bendermasteringstudio wrote:
Thanks, it's that on some music I think it sounds better to go straight from 96kHz to 44.1kHz mp3.
 

It's still getting SRC'd either way, and the SRC is most likely occuring prior to the actual MP3 encoding.  So this suggest you prefer the sound of your MP3 encoders SRC to whatever you typically use.  For me I always SRC (using iZotope 64bit SRC) to the designated sample rate prior to doing MP3 encoding (I like and use the LAME MP3 encoder).  I get the best results this way as opposed to allowing my MP3 encoder to handle the SRC.  Not that it does a bad job of SRC but iZotope's 64bit SRC does a better job to my ears. 

Quote    Reply   
avatar

bendermasteringstudio

Silverado

Posts: 220 Member Since:21/02/2011

#29 [url]

Aug 20 14 1:37 PM

aleatoric wrote:
bendermasteringstudio wrote:
Thanks, it's that on some music I think it sounds better to go straight from 96kHz to 44.1kHz mp3.
 

It's still getting SRC'd either way, and the SRC is most likely occuring prior to the actual MP3 encoding.  So this suggest you prefer the sound of your MP3 encoders SRC to whatever you typically use.  For me I always SRC (using iZotope 64bit SRC) to the designated sample rate prior to doing MP3 encoding (I like and use the LAME MP3 encoder).  I get the best results this way as opposed to allowing my MP3 encoder to handle the SRC.  Not that it does a bad job of SRC but iZotope's 64bit SRC does a better job to my ears. 


Indeed, on same music I do prefer encoding directly, but I rarely do 44.1kHz mp3. I only do them if a client reports some error or doubts about 48kHz mp3, which happened only a couple of times.

Just run some tests today and its the 48kHz mp3 that does sound better than the 44.1kHz ones, not the fact that I was feeding a 96 source kHz. I've tried downsampling it to 48kHz with iZotope before encoding and it is slightly better than going direct through LAME's encoder. So I'll be doing that from now on...

In any case, I rechecked and I still think going from a 96kHz source to a 48 kHz mp3 sounds way better than doing a 44.1kHz mp3 from a 44.1kHz source, regardless of the SRC used.

Quote    Reply   
avatar

biigniick

Silverado

Posts: 157 Member Since:16/02/2011

#30 [url]

Aug 20 14 5:56 PM

bendermasteringstudio wrote:
 


But have you AB both options? Going straight from a 96kHz wav to a 48khz mp3 using

 

no, i haven't.  the only AB test i've done is SRC to whatever final rate then convert to mp3 or convert to mp3 to same final sample rate.  i haven't setup the test with two identical DAC's at the different rates through a passive AB switch.  interesting test though...  maybe i'll do that sometime soon

- nick

Quote    Reply   
avatar

biigniick

Silverado

Posts: 157 Member Since:16/02/2011

#31 [url]

Aug 20 14 6:04 PM

bendermasteringstudio wrote:

Indeed, on same music I do prefer encoding directly, but I rarely do 44.1kHz mp3. I only do them if a client reports some error or doubts about 48kHz mp3, which happened only a couple of times.

Just run some tests today and its the 48kHz mp3 that does sound better than the 44.1kHz ones, not the fact that I was feeding a 96 source kHz. I've tried downsampling it to 48kHz with iZotope before encoding and it is slightly better than going direct through LAME's encoder. So I'll be doing that from now on...

In any case, I rechecked and I still think going from a 96kHz source to a 48 kHz mp3 sounds way better than doing a 44.1kHz mp3 from a 44.1kHz source, regardless of the SRC used.

awesome to get some good feedback on this.  how did you setup your test?  i'm always interested in other's test setups.

as for the 48k mp3 sounding better than the 44.1k, that seems counterintuitive to me...  with a constant bit rate mp3 and a higher sample rate, wouldn't that mean less bits per sample and reduced resolution?  i wonder where the cost/reward balance on this in regard to constant bitrate mp3 and sample rate.

i rember a time when i accitentaly encoded 96kHz mp3's that sounded like garbage!  they sounded very garbled and lots of artifacts, like a bitcrusher.  i remember thinking i was halfing the resolution of every sample to double the number of samples in the file.

i'm really enjoying this great conversation on this thread!  keep the great ideas coming :-)

- nick

Quote    Reply   
avatar

ShawnJHatfield

Silverado

Posts: 187 Member Since:19/05/2011

#32 [url]

Aug 21 14 1:12 AM

biigniick wrote:

with a constant bit rate mp3 and a higher sample rate, wouldn't that mean less bits per sample and reduced resolution?  i wonder where the cost/reward balance on this in regard to constant bitrate mp3 and sample rate.

Interesting point and something I hadn't considered until now.

Shawn Hatfield www.audibleoddities.com

Quote    Reply   
avatar

biigniick

Silverado

Posts: 157 Member Since:16/02/2011

#33 [url]

Aug 21 14 12:55 PM

twerk wrote:

biigniick wrote:

with a constant bit rate mp3 and a higher sample rate, wouldn't that mean less bits per sample and reduced resolution?  i wonder where the cost/reward balance on this in regard to constant bitrate mp3 and sample rate.

Interesting point and something I hadn't considered until now.

i ran into this on accident, like i said before, with 96k sample rate.  it made sense to me in my head ;-P

- nick
 

Quote    Reply   
avatar

chrisj

Gold Finger

Posts: 938 Member Since:22/02/2011

#34 [url]

Aug 21 14 3:25 PM

That's a REALLY good point. lossy encoders have to throw away most of what's up near Nyquist anyhow. I guess it might vary depending on the mix you're running into it: past a certain point for any frequency the encoder will discard the information. That noise floor level varies with frequency and is calculated accordingly.

If you're feeding the encoder a treble-heavy mix, 48K is absolutely the worst thing you could do for lossy encoding. You'd want it to give up filling in data somewhere short of the threshold of hearing, not go on adding more information very expensively that you can't hear.

Chris Johnson, airwindows.com

Quote    Reply   
avatar

reynaud

Silverado

Posts: 194 Member Since:13/07/2011

#35 [url]

Sep 11 14 9:44 AM

bendermasteringstudio wrote:
I've been looking for a decent batch processor to handle audio files. Something like what's built-in in iZotope RX, which I already have, but that could also handle lossy files and maybe even export to FLAC.
 

Rx4 now offers FLAC and OGG exports.

Quote    Reply   
Add Reply

Quick Reply

bbcode help