44.1kHz, 64Kbps, CBR mono encoding question

[Stevo Brock-2]

Hi list,

We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,  
CBR mono mp3 file and are having trouble matching the quality of the  
same file produced by ProTools.

Currently we are using the following settings:
Source file 44.1kHz
64Kbps
CBR
Stereo Mode Mono
Quality: 9

which gets us pretty close, but there are still some slight audible  
artifacts.  The ProTools file has no audible artifacts at all.

Interestingly, as quality setting is increased, the output actually  
gets worse.

We don't know anything about the advanced acoustic settings.  Would  
that help our situation?

Sample files are:
-original WAVE: http://monkey-tools.com/downloads/01_11a.wav
-encoded at lowest quality: http://monkey-tools.com/downloads/01_11a% 
20Q9.mp3
-encoded at highest quality: http://monkey-tools.com/downloads/01_11a% 
20Q0.mp3
-ProTools encoded: http://monkey-tools.com/downloads/01_11a_Protools.mp3

Any help or suggestions would be greatly appreciated.

-Stevo Brock
  Head of Development
  Monkey Tools
  www.monkey-tools.com




_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Manoj Bist]

Perhaps you should use -q 2 (this is the default).

-q 0 is the highest quality and -q 9 is the lowest quality.

On 10/19/07, Stevo Brock <[hidden email]> wrote:

>
> Hi list,
>
> We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,
> CBR mono mp3 file and are having trouble matching the quality of the
> same file produced by ProTools.
>
> Currently we are using the following settings:
> Source file 44.1kHz
> 64Kbps
> CBR
> Stereo Mode Mono
> Quality: 9
>
> which gets us pretty close, but there are still some slight audible
> artifacts.  The ProTools file has no audible artifacts at all.
>
> Interestingly, as quality setting is increased, the output actually
> gets worse.
>
> We don't know anything about the advanced acoustic settings.  Would
> that help our situation?
>
> Sample files are:
> -original WAVE: http://monkey-tools.com/downloads/01_11a.wav
> -encoded at lowest quality: http://monkey-tools.com/downloads/01_11a%
> 20Q9.mp3
> -encoded at highest quality: http://monkey-tools.com/downloads/01_11a%
> 20Q0.mp3
> -ProTools encoded: http://monkey-tools.com/downloads/01_11a_Protools.mp3
>
> Any help or suggestions would be greatly appreciated.
>
> -Stevo Brock
>   Head of Development
>   Monkey Tools
>   www.monkey-tools.com
>
>
>
>
> _______________________________________________
> mp3encoder mailing list
> [hidden email]
> https://minnie.tuhs.org/mailman/listinfo/mp3encoder
>

_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Manoj Bist]

Can you include the lame command line that you are using?
By default lame assumes stereo mode, for mono you need to use option '-m m'.


On 10/19/07, Manoj Bist <[hidden email]> wrote:

>
> Perhaps you should use -q 2 (this is the default).
>
> -q 0 is the highest quality and -q 9 is the lowest quality.
>
> On 10/19/07, Stevo Brock < [hidden email]> wrote:
> >
> > Hi list,
> >
> > We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,
> > CBR mono mp3 file and are having trouble matching the quality of the
> > same file produced by ProTools.
> >
> > Currently we are using the following settings:
> > Source file 44.1kHz
> > 64Kbps
> > CBR
> > Stereo Mode Mono
> > Quality: 9
> >
> > which gets us pretty close, but there are still some slight audible
> > artifacts.  The ProTools file has no audible artifacts at all.
> >
> > Interestingly, as quality setting is increased, the output actually
> > gets worse.
> >
> > We don't know anything about the advanced acoustic settings.  Would
> > that help our situation?
> >
> > Sample files are:
> > -original WAVE: http://monkey-tools.com/downloads/01_11a.wav
> > -encoded at lowest quality: http://monkey-tools.com/downloads/01_11a%<http://monkey-tools.com/downloads/01_11a%25>
> > 20Q9.mp3
> > -encoded at highest quality: http://monkey-tools.com/downloads/01_11a%<http://monkey-tools.com/downloads/01_11a%25>
> > 20Q0.mp3
> > -ProTools encoded: http://monkey-tools.com/downloads/01_11a_Protools.mp3
> >
> >
> > Any help or suggestions would be greatly appreciated.
> >
> > -Stevo Brock
> >   Head of Development
> >   Monkey Tools
> >   www.monkey-tools.com
> >
> >
> >
> >
> > _______________________________________________
> > mp3encoder mailing list
> > [hidden email]
> > https://minnie.tuhs.org/mailman/listinfo/mp3encoder
> >
>
>

_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Bill Kincaid]

Interesting.  I looked at these encodings with our analysis tool and the
thing that stands out in comparing the Q0 to Q9 encodings is that the Q9
does no amplifications at all (scalefactors all zero).  The ProTools
encoding also does less amplification (no scalefac_scale use) and manages
the bit reservoir better.

This is probably academic to you unless you are an encoder developer!

One thing you could try is downsampling to 22.050.  For voice recording that
should be plenty of bandwidth and it will let the encoder use the bits more
efficiently.

Cheers,
-Bill

> From: Stevo Brock <[hidden email]>
> Reply-To: MP3 encoders development list <[hidden email]>
> Date: Fri, 19 Oct 2007 12:35:25 -0700
> To: MP3 encoders development list <[hidden email]>
> Subject: [mp3encoder] 44.1kHz, 64Kbps, CBR mono encoding question
>
> Hi list,
>
> We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,
> CBR mono mp3 file and are having trouble matching the quality of the
> same file produced by ProTools.
>
> Currently we are using the following settings:
> Source file 44.1kHz
> 64Kbps
> CBR
> Stereo Mode Mono
> Quality: 9
>
> which gets us pretty close, but there are still some slight audible
> artifacts.  The ProTools file has no audible artifacts at all.
>
> Interestingly, as quality setting is increased, the output actually
> gets worse.
>
> We don't know anything about the advanced acoustic settings.  Would
> that help our situation?
>
> Sample files are:
> -original WAVE: http://monkey-tools.com/downloads/01_11a.wav
> -encoded at lowest quality: http://monkey-tools.com/downloads/01_11a%
> 20Q9.mp3
> -encoded at highest quality: http://monkey-tools.com/downloads/01_11a%
> 20Q0.mp3
> -ProTools encoded: http://monkey-tools.com/downloads/01_11a_Protools.mp3
>
> Any help or suggestions would be greatly appreciated.
>
> -Stevo Brock
>   Head of Development
>   Monkey Tools
>   www.monkey-tools.com
>
>
>
>
> _______________________________________________
> mp3encoder mailing list
> [hidden email]
> https://minnie.tuhs.org/mailman/listinfo/mp3encoder

_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Jaroslav Lukesh]

----- Original Message -----
From: "Stevo Brock" <[hidden email]>

> We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,
> CBR mono mp3 file and are having trouble matching the quality of the
> same file produced by ProTools.
>
> Currently we are using the following settings:
> Source file 44.1kHz
> 64Kbps
> CBR
> Stereo Mode Mono
> Quality: 9
>
> which gets us pretty close, but there are still some slight audible
> artifacts.  The ProTools file has no audible artifacts at all.
>

Lame is no good encoder for bitrates <160 kbps CBR or <130 kbps ABR.
Fastencc (as reference highest possible quality encoder) which is part of
protools (if you mean ten-thousands$$$ musical HW&SW of course), is the best
for low bitrates as inet audio streaming etc.

Lame needs mode parameters tuning, as move ATH, different ATH curves etc.
And of course, fastencc does not use MDCT at highest quality mode, but
different algorithm, which gave not only spectral waves, but it relys on
phase shifts too.

Now I hear all audio academics with long titles before and after their
names, that ear does not recognize phase shifts... Yes, of course, ear does
not recognize phaseshifts at sinusoidal signal, persist from bigbang to
collapse of universe. But music is DYNAMIC with transients and here ear
should hear tones which is in some cases more than 30 dB under popular audio
"0dB" curves. And transients without original phase shifts are not right
formed transients, but something totally different. ANd there your ear and
brain does not recognize tones as with original signal.

You should make some tests for that: you should need very good audio
equipment (good <> astronomical price!), rather studio grade with even worse
parameters than Hi-End with 0,000000001% sinusoidal distortion. Now you can
listen for example R&B with deep plucked bass sounds at levels about 30dB.
Yeah, you should hear these deep bass sounds! So public theory about ATH
curves lies for that.

Studio grade equipments are constructed for transient signal which does not
have easy to measurable parameters, if any. But it have maximum sound
reality. Noise and sinusoidal distortion are not the main parameter. Hi-End
grade audio are constructed to reproduce sinusoidal signals which have nice
easy to measure parameters, but it doesnot reproduce natural sound as is.

If you want high quality, use studio equipment and does not look at their
technical parameters. Total output S/N of mixer like 80dB, THD about 0,01%
is normal... Yes, but it sounds BETTER than any Hi-End.

Regards, JL.



_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Stevo Brock-2]

And in fact that is what is interesting.  -q 9 gives the best  
results.  -q0 creates strange artifacts in the file.

-Stevo


On Oct 19, 2007, at 12:50 PM, Manoj Bist wrote:

> Perhaps you should use -q 2 (this is the default).
>
> -q 0 is the highest quality and -q 9 is the lowest quality.
>
> On 10/19/07, Stevo Brock <[hidden email]> wrote:
>>
>> Hi list,
>>
>> We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,
>> CBR mono mp3 file and are having trouble matching the quality of the
>> same file produced by ProTools.
>>
>> Currently we are using the following settings:
>> Source file 44.1kHz
>> 64Kbps
>> CBR
>> Stereo Mode Mono
>> Quality: 9
>>
>> which gets us pretty close, but there are still some slight audible
>> artifacts.  The ProTools file has no audible artifacts at all.
>>
>> Interestingly, as quality setting is increased, the output actually
>> gets worse.
>>
>> We don't know anything about the advanced acoustic settings.  Would
>> that help our situation?
>>
>> Sample files are:
>> -original WAVE: http://monkey-tools.com/downloads/01_11a.wav
>> -encoded at lowest quality: http://monkey-tools.com/downloads/01_11a%
>> 20Q9.mp3
>> -encoded at highest quality: http://monkey-tools.com/downloads/ 
>> 01_11a%
>> 20Q0.mp3
>> -ProTools encoded: http://monkey-tools.com/downloads/ 
>> 01_11a_Protools.mp3
>>
>> Any help or suggestions would be greatly appreciated.
>>
>> -Stevo Brock
>>   Head of Development
>>   Monkey Tools
>>   www.monkey-tools.com
>>
>>
>>
>>
>> _______________________________________________
>> mp3encoder mailing list
>> [hidden email]
>> https://minnie.tuhs.org/mailman/listinfo/mp3encoder
>>
> _______________________________________________
> mp3encoder mailing list
> [hidden email]
> https://minnie.tuhs.org/mailman/listinfo/mp3encoder
>

_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Robert Hegemann]

Hello Stevo,

Am Freitag, 19. Oktober 2007 21:35 schrieb Stevo Brock:
> Hi list,
>
> We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,
> CBR mono mp3 file and are having trouble matching the quality of the
> same file produced by ProTools.

how about this, it uses similar lowpass as ProTools:

 lame --lowpass 13.6 01_11a.wav maybe_1.mp3

or for a VBR file (*)

 lame -V5 01_11a.wav maybe_2.mp3

or for another VBR file (*)

 lame -V5 --vbr-new 01_11a.wav maybe_3.mp3



(*) If it has to be CBR, you may try this too:

 lame -V5 -b64 -F 01_11a.wav maybe_4.mp3

> Currently we are using the following settings:
> Source file 44.1kHz
> 64Kbps
> CBR
> Stereo Mode Mono
> Quality: 9
>
> which gets us pretty close, but there are still some slight audible
> artifacts.  The ProTools file has no audible artifacts at all.
>
> Interestingly, as quality setting is increased, the output actually
> gets worse.
>
> We don't know anything about the advanced acoustic settings.  Would
> that help our situation?
>
> Sample files are:
> -original WAVE: http://monkey-tools.com/downloads/01_11a.wav
> -encoded at lowest quality: http://monkey-tools.com/downloads/01_11a%
> 20Q9.mp3
> -encoded at highest quality: http://monkey-tools.com/downloads/01_11a%
> 20Q0.mp3
> -ProTools encoded: http://monkey-tools.com/downloads/01_11a_Protools.mp3
>
> Any help or suggestions would be greatly appreciated.
>
> -Stevo Brock
>   Head of Development
>   Monkey Tools
>   www.monkey-tools.com

Ciao Robert


_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Stevo Brock-2]

Thanks for that info.  It is interesting.  I have tried lowering the  
sample rate, upping the bitrate, and changing to VBR, all with great  
results.  Unfortunately, the customer we are working with has hard  
client requirements and isn't able to budge.

-Stevo Brock
  Head of Development
  Monkey Tools
  www.monkey-tools.com


On Oct 19, 2007, at 1:17 PM, Bill Kincaid wrote:

> Interesting.  I looked at these encodings with our analysis tool  
> and the
> thing that stands out in comparing the Q0 to Q9 encodings is that  
> the Q9
> does no amplifications at all (scalefactors all zero).  The ProTools
> encoding also does less amplification (no scalefac_scale use) and  
> manages
> the bit reservoir better.
>
> This is probably academic to you unless you are an encoder developer!
>
> One thing you could try is downsampling to 22.050.  For voice  
> recording that
> should be plenty of bandwidth and it will let the encoder use the  
> bits more
> efficiently.
>
> Cheers,
> -Bill
>
>
>> From: Stevo Brock <[hidden email]>
>> Reply-To: MP3 encoders development list <[hidden email]>
>> Date: Fri, 19 Oct 2007 12:35:25 -0700
>> To: MP3 encoders development list <[hidden email]>
>> Subject: [mp3encoder] 44.1kHz, 64Kbps, CBR mono encoding question
>>
>> Hi list,
>>
>> We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,
>> CBR mono mp3 file and are having trouble matching the quality of the
>> same file produced by ProTools.
>>
>> Currently we are using the following settings:
>> Source file 44.1kHz
>> 64Kbps
>> CBR
>> Stereo Mode Mono
>> Quality: 9
>>
>> which gets us pretty close, but there are still some slight audible
>> artifacts.  The ProTools file has no audible artifacts at all.
>>
>> Interestingly, as quality setting is increased, the output actually
>> gets worse.
>>
>> We don't know anything about the advanced acoustic settings.  Would
>> that help our situation?
>>
>> Sample files are:
>> -original WAVE: http://monkey-tools.com/downloads/01_11a.wav
>> -encoded at lowest quality: http://monkey-tools.com/downloads/01_11a%
>> 20Q9.mp3
>> -encoded at highest quality: http://monkey-tools.com/downloads/ 
>> 01_11a%
>> 20Q0.mp3
>> -ProTools encoded: http://monkey-tools.com/downloads/ 
>> 01_11a_Protools.mp3
>>
>> Any help or suggestions would be greatly appreciated.
>>
>> -Stevo Brock
>>   Head of Development
>>   Monkey Tools
>>   www.monkey-tools.com
>>
>>
>>
>>
>> _______________________________________________
>> mp3encoder mailing list
>> [hidden email]
>> https://minnie.tuhs.org/mailman/listinfo/mp3encoder
>
>
> _______________________________________________
> mp3encoder mailing list
> [hidden email]
> https://minnie.tuhs.org/mailman/listinfo/mp3encoder
>

_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Stevo Brock-2]

We're not using the command line interface, but the API.  To answer  
your question, in this case, we are calling:

                        status = lame_set_mode(mPrivateData->mLameFlags, MONO);

-Stevo Brock
  Head of Development
  Monkey Tools
  www.monkey-tools.com




On Oct 19, 2007, at 12:54 PM, Manoj Bist wrote:

> Can you include the lame command line that you are using?
> By default lame assumes stereo mode, for mono you need to use  
> option '-m m'.
>
>
> On 10/19/07, Manoj Bist <[hidden email]> wrote:
>>
>> Perhaps you should use -q 2 (this is the default).
>>
>> -q 0 is the highest quality and -q 9 is the lowest quality.
>>
>> On 10/19/07, Stevo Brock < [hidden email]> wrote:
>>>
>>> Hi list,
>>>
>>> We are trying to use LAME to encode a WAVE file to a 44.1kHz,  
>>> 64Kbps,
>>> CBR mono mp3 file and are having trouble matching the quality of the
>>> same file produced by ProTools.
>>>
>>> Currently we are using the following settings:
>>> Source file 44.1kHz
>>> 64Kbps
>>> CBR
>>> Stereo Mode Mono
>>> Quality: 9
>>>
>>> which gets us pretty close, but there are still some slight audible
>>> artifacts.  The ProTools file has no audible artifacts at all.
>>>
>>> Interestingly, as quality setting is increased, the output actually
>>> gets worse.
>>>
>>> We don't know anything about the advanced acoustic settings.  Would
>>> that help our situation?
>>>
>>> Sample files are:
>>> -original WAVE: http://monkey-tools.com/downloads/01_11a.wav
>>> -encoded at lowest quality: http://monkey-tools.com/downloads/ 
>>> 01_11a%<http://monkey-tools.com/downloads/01_11a%25>
>>> 20Q9.mp3
>>> -encoded at highest quality: http://monkey-tools.com/downloads/ 
>>> 01_11a%<http://monkey-tools.com/downloads/01_11a%25>
>>> 20Q0.mp3
>>> -ProTools encoded: http://monkey-tools.com/downloads/ 
>>> 01_11a_Protools.mp3
>>>
>>>
>>> Any help or suggestions would be greatly appreciated.
>>>
>>> -Stevo Brock
>>>   Head of Development
>>>   Monkey Tools
>>>   www.monkey-tools.com
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> mp3encoder mailing list
>>> [hidden email]
>>> https://minnie.tuhs.org/mailman/listinfo/mp3encoder
>>>
>>
>>
> _______________________________________________
> mp3encoder mailing list
> [hidden email]
> https://minnie.tuhs.org/mailman/listinfo/mp3encoder
>

_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder

[Stevo Brock-2]

Hi Jaroslav,

You mentioned the mode parameter tuning of LAME.  That's something I  
know very little about.  Is there any overview advice you can give  
about that?

-Stevo Brock
  Head of Development
  Monkey Tools
  www.monkey-tools.com




On Oct 19, 2007, at 1:57 PM, Jaroslav Lukesh wrote:

> ----- Original Message -----
> From: "Stevo Brock" <[hidden email]>
>> We are trying to use LAME to encode a WAVE file to a 44.1kHz, 64Kbps,
>> CBR mono mp3 file and are having trouble matching the quality of the
>> same file produced by ProTools.
>>
>> Currently we are using the following settings:
>> Source file 44.1kHz
>> 64Kbps
>> CBR
>> Stereo Mode Mono
>> Quality: 9
>>
>> which gets us pretty close, but there are still some slight audible
>> artifacts.  The ProTools file has no audible artifacts at all.
>>
>
> Lame is no good encoder for bitrates <160 kbps CBR or <130 kbps ABR.
> Fastencc (as reference highest possible quality encoder) which is  
> part of
> protools (if you mean ten-thousands$$$ musical HW&SW of course), is  
> the best
> for low bitrates as inet audio streaming etc.
>
> Lame needs mode parameters tuning, as move ATH, different ATH  
> curves etc.
> And of course, fastencc does not use MDCT at highest quality mode, but
> different algorithm, which gave not only spectral waves, but it  
> relys on
> phase shifts too.
>
> Now I hear all audio academics with long titles before and after their
> names, that ear does not recognize phase shifts... Yes, of course,  
> ear does
> not recognize phaseshifts at sinusoidal signal, persist from  
> bigbang to
> collapse of universe. But music is DYNAMIC with transients and here  
> ear
> should hear tones which is in some cases more than 30 dB under  
> popular audio
> "0dB" curves. And transients without original phase shifts are not  
> right
> formed transients, but something totally different. ANd there your  
> ear and
> brain does not recognize tones as with original signal.
>
> You should make some tests for that: you should need very good audio
> equipment (good <> astronomical price!), rather studio grade with  
> even worse
> parameters than Hi-End with 0,000000001% sinusoidal distortion. Now  
> you can
> listen for example R&B with deep plucked bass sounds at levels  
> about 30dB.
> Yeah, you should hear these deep bass sounds! So public theory  
> about ATH
> curves lies for that.
>
> Studio grade equipments are constructed for transient signal which  
> does not
> have easy to measurable parameters, if any. But it have maximum sound
> reality. Noise and sinusoidal distortion are not the main  
> parameter. Hi-End
> grade audio are constructed to reproduce sinusoidal signals which  
> have nice
> easy to measure parameters, but it doesnot reproduce natural sound  
> as is.
>
> If you want high quality, use studio equipment and does not look at  
> their
> technical parameters. Total output S/N of mixer like 80dB, THD  
> about 0,01%
> is normal... Yes, but it sounds BETTER than any Hi-End.
>
> Regards, JL.
>
>
>
> _______________________________________________
> mp3encoder mailing list
> [hidden email]
> https://minnie.tuhs.org/mailman/listinfo/mp3encoder
>

_______________________________________________
mp3encoder mailing list
[hidden email]
https://minnie.tuhs.org/mailman/listinfo/mp3encoder