Normalize Codec - Peak to Peak not matching output

[Treelady]

Let me know if I'm doing something wrong. But I can't seem to get my Normalize function to do what I want it to do.


Release 17.7 [64-bit]
OS Windows 10 Pro 64

Objective - after the conversion, (24-bit 96kHz WAV to 16-bit 44.1kHz WAV) I want the sample peak of a file to be a given value. For this example, I'll say -2.0 dBFS

Settings used in Encoder
DSP Effect
Volume Normalize
TYPE: Peak to Peak
Desired Volume: -2
Maximum Amplification x8
Info: Non-Live DSP Effect

However, after the encoding, the statistics do not reflect the intended result. I understand True Peak Levels will be higher than peak levels, but is a 0.6 db difference within reasonable variance? In general, I encounter less than .3 db difference between peak and true peak when using Weiss Saracon.


Statistical testing with several applications reports the same result: Max TPL is around -1.4

MAAT DR2 Offline
Max. TP: -1.39

YOULEAN LOUDNESS METER
True Peak Max: -1.4 db

Acon Digital Acoustica

Maximum True Peak Level: -1.3 db

Any thoughts would be appreciated. Thank you in advance.

[Spoon]

Re: Normalize Codec - Peak to Peak not matching output

You have to use the DSP Effects:

Resample
Bit depth
Volume Normalize

otherwise the reduction to 16 bit happens after the volume normalize (which would be done at 24 bit 96KHz)

[Treelady]

Re: Normalize Codec - Peak to Peak not matching output

You have to use the DSP Effects:

Resample
Bit depth
Volume Normalize

otherwise the reduction to 16 bit happens after the volume normalize (which would be done at 24 bit 96KHz)

I'm not getting any difference in doing it the way I originally set up the job vs what you're suggesting.

I'll stick with my example of going 24/96 to 16/44. Sometimes examples help keep my questions clearer.

So the order of operations is:

Input File (24/96) > Any DSP effects (normalization)> Conversion (16/44)

You are suggesting I manually force the conversion through DSP:

Input File (24/96) > DSP chain: Resample (24/44) > Bit depth (16/44) > Volume Normalize (16/44) >

At this point what happens to the desired output parameters set in the dialogue box requested output =
(16/44)?

Does the application look at the state of the file (16/44) and determine that the bit depth and SR are already at the desired output (16/44)? Or does it convert it again to the same format it already is? (16/44 to 16/44)

I'm sorry to ask these kinds of questions, but I made the assumption that the DSP effects happened AFTER the conversion.

I did two tests after I completed the conversions:

1. Basic Statistics by MAAT DR2

File Name Max. TPL
Convert specified then DSP -1.97
Forcing with DSP -1.97

2. Null Test

these null to 0

[Spoon]

Re: Normalize Codec - Peak to Peak not matching output

If the bit depth/sample rate is already 16/44 then the wave encoder should not further process it, however you can set to 'as source' to be sure, in this instance source is how the data arrives so would be 16/44

[Treelady]

Re: Normalize Codec - Peak to Peak not matching output

If the bit depth/sample rate is already 16/44 then the wave encoder should not further process it, however you can set to 'as source' to be sure, in this instance source is how the data arrives so would be 16/44

Thank you. This has been very helpful!