3 Oct 2009 11:40

### Passband question

```In SRC page it's referred:
"The current best converter provides a signal-to-noise ratio of 145dB
with -3dB passband extending from DC to 96% of the theoretical best
bandwidth for a given pair of input and output sample rates"
From my understanding, when we resample to a lower frequency, the passband
of the original signal is reduced, but if we resample to a higher frequency,
the passband of the original signal could be kept intact if the new sample
rate frequency is high enough.

Example:
a)48000Hz -> 44100Hz
Input [0,24000] ; Output [0, 21168]

b)48000Hz -> 96000Hz
Input [0,24000] ; Output [0, 46080]

Is this correct, or is the attenuation performed on the input signal,
independently of the output frequency selected, hence the output never
having any "valid" frequency above 23040Hz?

Thanks.

------------------------------------------------------------
Secret Rabbit Code (aka libsamplerate) Mailing List
To unsubscribe:
<mailto:src-request@...?Subject=unsubscribe>

```
3 Oct 2009 23:28

### Re: Passband question

```Hi Samunel,

First off, if you want to start a new topic of conversation, please do not
reply to an existing thread. Insteand, compose a new message and enter the
list address in the To: field.

Samuel Grave wrote:

> In SRC page it's referred:
> "The current best converter provides a signal-to-noise ratio of 145dB
> with -3dB passband extending from DC to 96% of the theoretical best
> bandwidth for a given pair of input and output sample rates"
> From my understanding, when we resample to a lower frequency, the passband
> of the original signal is reduced, but if we resample to a higher frequency,
> the passband of the original signal could be kept intact if the new sample
> rate frequency is high enough.
>
> Example:
> a)48000Hz -> 44100Hz
>     Input [0,24000] ; Output [0, 21168]
>
> b)48000Hz -> 96000Hz
>     Input [0,24000] ; Output [0, 46080]

I think you meant 23040 there?

> Is this correct, or is the attenuation performed on the input signal,
> independently of the output frequency selected, hence the output never
> having any "valid" frequency above 23040Hz?

```

4 Oct 2009 03:21

### Re: Passband question

```Hi Erik,
Sorry for that. I've changed the subject and thought that was enough...

>> b)48000Hz -> 96000Hz
>>     Input [0,24000] ; Output [0, 46080]
>
> I think you meant 23040 there?
No, I really meant 46080. I was considering the possibility that resampling
to a higher frequency could increase the -3dB frequency, that's why I
referred 46080.
In the mean time, I've decided to perform some tests and have found the

> No quite. The attenution is 3db at 23040Hz increases to the full
> attenuation (dependent on which converter is used) at 24000.
I realize now that my question did not reflected what I wanted to ask...
Sorry, but thanks for answering it anyway.

Best Regards.

------------------------------------------------------------
Secret Rabbit Code (aka libsamplerate) Mailing List
To unsubscribe:
<mailto:src-request@...?Subject=unsubscribe>

```
4 Oct 2009 06:16

### Re: Passband question

```Samuel Grave wrote:

> No, I really meant 46080. I was considering the possibility that resampling
> to a higher frequency could increase the -3dB frequency, that's why I
> referred 46080.

After upsampling the theoretical bandwith is indeed 48000Hz, but there
because the source signal had no frequency components above 24000, the
output signal should have no frequency components above 24000Hz. There
will however be resampling artifacts, but they will be below the noise
floor (ie 144dB for BEST converter).

Cheers,
Erik
--

--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
------------------------------------------------------------
Secret Rabbit Code (aka libsamplerate) Mailing List
To unsubscribe:
<mailto:src-request@...?Subject=unsubscribe>

```
4 Oct 2009 23:24

### Re: Passband question

```Samuel Grave wrote:

> This is the only part that it's not clear enough for me... I know that
> outside half the input sample rate it would only be noise, but is it
> dangerous to the equipment?

No, not dangerous to the equipment, but very dangerous for sound
quality .

> Why not giving us the option of lowpassing at half the output sample rate,
> or to a sample rate that we can specify, so we could be able to preserve the
> input without any attenuation?

The design of libsamplerate is as close to optimal as I can get it. Giving
users options to change parameters can only make for a worse sample rate
converter.

Erik
--

--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
------------------------------------------------------------
Secret Rabbit Code (aka libsamplerate) Mailing List
To unsubscribe:
<mailto:src-request@...?Subject=unsubscribe>

```
5 Oct 2009 01:35

### Re: Passband question

```On Sun, Oct 4, 2009 at 5:24 PM, Erik de Castro Lopo <erikd@...> wrote:
> The design of libsamplerate is as close to optimal as I can get it. Giving
> users options to change parameters can only make for a worse sample rate
> converter.

This is especially true because the difference in sound quality
between a very good SRC and a somewhat poor one is quite subtle. It's
very likely that if a random user twiddled the settings they'd turn it
into a poor quality SRC and not notice.
------------------------------------------------------------
Secret Rabbit Code (aka libsamplerate) Mailing List
To unsubscribe:
<mailto:src-request@...?Subject=unsubscribe>

```
5 Oct 2009 02:00

### Re: Passband question

```> The design of libsamplerate is as close to optimal as I can get it. Giving
> users options to change parameters can only make for a worse sample rate
> converter.
I understand that.
Just for curiosity, I've messed a little with libsamplerate code trying to
disable the low pass filtering, but I think I was not able to do it...
With my first try I had a horrible 23000Hz sin signal with lots of
artifacts. With my second try, I was able to decrease the attenuation and
keep the same wave format that I got with the low pass filter, but I haven't
tryed how it sounded. Now, since there would be no problem in damaging the
equipment, I think I will give it a try just for fun...

Best Regards,
Samuel

------------------------------------------------------------
Secret Rabbit Code (aka libsamplerate) Mailing List
To unsubscribe:
<mailto:src-request@...?Subject=unsubscribe>

```
28 Oct 2009 22:31

### List archive?

```Hello everyone!
A couple of quick questions.

Is there an archive for this list? I would like to see if an issue I'm
having with SRC has already been addressed before I post my problem.

Hopefull I wont need to post again, but if so, what are the guidelines
on attachments? I can better illustrate my SRC problem with a few
screeshots and some source code; should I attach them or link to them
elsewhere.

Thanks!

----
Troy
```
28 Oct 2009 22:31

### List archive?

```Hello everyone!
A couple of quick questions.

Is there an archive for this list? I would like to see if an issue I'm
having with SRC has already been addressed before I post my problem.

Hopefull I wont need to post again, but if so, what are the guidelines
on attachments? I can better illustrate my SRC problem with a few
screeshots and some source code; should I attach them or link to them
elsewhere.

Thanks!

----
Troy
```
28 Oct 2009 23:13

### Re: List archive?

```Troy Watson wrote:

> Is there an archive for this list?

Unfortunately no.

> Hopefull I wont need to post again, but if so, what are the guidelines
> on attachments? I can better illustrate my SRC problem with a few
> screeshots and some source code; should I attach them or link to them
> elsewhere.

Link them elsewhere. The list doesn't accept attachments.

Erik
--

--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
```

Gmane