Topic: Converting Sample Rate/Add Gain ???

[T.J.]

okay here's the deal (i'm using wavelab 5.0):

-i have a wav file that is 16/48.
-i wanted to add some gain and checked for the peak. it was -1.238dB so i added 1.2dB of gain.
-i checked the peak again and it was -0.038, which makes sense.
-then i converted the sample rate from 48 to 44.1
-after converting to 44.1 i checked the peak and expected to see -0.038dB. instead the peak is -2.676.

my question is how does converting the sample rate change the highest peak dB?

[John Kary]

When you're using a bad Sample Rate Converter (SRC).

Select Wavelab 5 from the pull-down list and compare it to the others: http://src.infinitewave.ca/

Most people use R8brain to do SRC, as it's free, and produces some of the best results after iZotope RX Advanced.
http://www.voxengo.com/product/r8brain/

[taper420]

holy crap.... the one I've been using (audio desk/digital performer) is horrible.
i wonder what quicktime looks like on the best setting. I imagine pretty bad as well.

anybody have a clean converter such as R8brain they would recommend for a mac?

[T.J.]

i've only been recording at a higher sample/bit rate for the past year. all of my recordings have been edited and then resampled/dithered. all of the editing i do is just adding fades at the beginning/end of the set and add some gain (usually around 4 to 8dB). i remember hearing it is best to do all editing with the highest sample/bit rate and then resample/dither.

i'm aware that all audio software has it's own programs that perform resampling. but i had no idea it would directly affect the peak levels of a wave file. plus i would assume wavelab would have a good program for resampling. shouldn't i want to add gain AFTER dither/resample is completed if i want the file to maintain a peak closer to zero?

the thing i'm mainly concerned about is how this affects the 16/44.1 wav. if i boost my levels close to zero and then resample/dither, causing my peak level to fall to around -2dB (referring to the example from the original post), what the point of even adding gain?

[John Kary]

The point is, it's not keeping at the same level because the SRC you're using is of poor quality.  If you use better SRC/Dither processes, this gain flux won't come into play at all.

You want to do all of your editing at highest bitrate/sample rate, save it in its native bitrate/samplerate, use r8brain to convert to 44.1, then dither.

[T.J.]

no totally hear you and +T for helping. i have heard about R8brain and may start using it after comparing the results to that of wavelab's. it seems alot of people on this board use wavelab to resample. to me, it seems pointless to add gain if you are using wavelab's SRC. the biggest question i have is how does this really affect the sound of the completed 16/44.1 fileset?

let me try to elaborate: when adding gain to a 24/48 file you increase the volume of the recording. after coverting the sample rate the gain is now decreased because of the SRC i am using. what happened to the boost in volume? it appears there isn't a drastic change in the actual wav when i look at it in wavelab. what happened to the volume boost applied to the file? when playing back this file, can this SRC cause an unwanted change in volume? sorry i'm not really able to explain myself better.

edit to say: is it safe to assume all files i have added gain to then resampled/dithered in wavelab have a lower peak dB level then when i started? is that boost in gain i applied to all my 24/48 not boosting the volume of the 16/44.1 files?

[Gordon]

tj


are you using the "stock" resample in wavelab or the resamper 192 plugin?  I now use the plugin but use to use both and have never noticed anything like this.  I'm gonna run a few test right now.

[Gordon]

ok


16/48 wave file


peak level is -8.945

I raised it +8


so the peak level was then -.945

resampled with the "stock" convert sample rate in wavelab.

checked the gain and it was now -1.067


changed but barely.  not nearly as drastic as your example.


same file done with the resampler 192 plugin in wavelab the resulting file ended up being -1.026

so still a slight difference I can live with. yours however is over 2.5 dbs.  not sure why.

[T.J.]

tj


are you using the "stock" resample in wavelab or the resamper 192 plugin?  I now use the plugin but use to use both and have never noticed anything like this.  I'm gonna run a few test right now.

i use the option uder tools > convert sample rate, so i guess it's the stock.

[T.J.]

i did another test i did at home i'll post when i get done work. but basically i added 4.5dB of gain, resampled using the "stock" SRC that changed the peak, then used the UV22HR to dither which also changed the peak. so i ended up with a file that only had about 3dB of gain added. not a big deal...but i didn't know about this before.

[Gordon]

just did another test with a 24/96 wave

peak level was  -2.564

raised gain + 2 to make it -.564


"stock" resample made it -.629

resampler 192 plugin made it -.572


very slight difference.  looks like it happens more in the 16 bit realm.


I then dithered both with uv22hr


"stock" resample and dither with uv22hr  came out to     -.63


resampler 192 plugin and dither with uv22hr came out to -.571


so I am seeing very, very slight difference.  again no idea why yours was 2.5 dbs or so.


the resampler 192 is in your master plugin drop down.  just render in place.

[T.J.]

i have to try the resampler 192 when i get home. thanks gordon! +T

[easy jim]

holy crap.... the one I've been using (audio desk/digital performer) is horrible.
i wonder what quicktime looks like on the best setting. I imagine pretty bad as well.

anybody have a clean converter such as R8brain they would recommend for a mac?

Well, they don't indicate in the test graphs which version of DP was used for the testing, so it is possibly a different algorithm/SRC between DP 5.1/AudioDesk 2.0 vs. older versions of DP (4._ or earlier, I'm guessing, is what they used) if that has changed with subsequent updates.

From all the formally trained engineers I know that are Mac folks, and some others who are more hobbyists like the average TS.com member, I understand Bias Peak Pro to be the 'standard' or 'best' software for sample rate conversion on the Mac (others, I expect, will likely differ here).  It's not cheap, though, at $499 (or $399 w/ educational discount).  I have not been able to hear a difference in some sloppy (crappy playback) comparisons I have heard of individual tracks, though I know that is not saying much.  AD/DP have been 'good enough' for me for a while, though SRC in both programs is excrusciatingly slow (more so in AD).

[wbrisette]

From all the formally trained engineers I know that are Mac folks, and some others who are more hobbyists like the average TS.com member, I understand Bias Peak Pro to be the 'standard' or 'best' software for sample rate conversion on the Mac (others, I expect, will likely differ here).

This use to be the case. However, there is a new sheriff in town and it goes by the name Izotope. Their Ozone 3 product (like their RX product) put most of the other player's to shame. Audiofile Engineering licensed the iZotope 64-bit SRC and put it into WaveEditor. That product has come a long way in a short time. They even allow you to do DDP with the basic version of the product and it cost $200. A steal compared to Bias. I still own Peak Pro, but I rarely find myself using it now that I've gotten the hang of WaveEditor.

Wayne

[easy jim]

From all the formally trained engineers I know that are Mac folks, and some others who are more hobbyists like the average TS.com member, I understand Bias Peak Pro to be the 'standard' or 'best' software for sample rate conversion on the Mac (others, I expect, will likely differ here).

This use to be the case. However, there is a new sheriff in town and it goes by the name Izotope. Their Ozone 3 product (like their RX product) put most of the other player's to shame. Audiofile Engineering licensed the iZotope 64-bit SRC and put it into WaveEditor. That product has come a long way in a short time. They even allow you to do DDP with the basic version of the product and it cost $200. A steal compared to Bias. I still own Peak Pro, but I rarely find myself using it now that I've gotten the hang of WaveEditor.

Wayne

Thanks Wayne.  +t  Now I'm curious to check out what WaveEditor has to offer.