DSPRelated.com
Forums

Theoretically Highest Quality of PCM Audio

Started by Radium November 9, 2003

Radium wrote:
> > What is the theoretically highest possible: > > 1. Frequency Response (range of frequencies) > > 2. SNR > > 3. Dynamic Range > > 4. SPL > > 5. Musical Pitch (highest acceptable frequency) > > of PCM audio?
There is no theoretical limit. However there are different standards which specify the performance, and the different hardware limitations. The today's state of art in the pro audio is ~24kHz, ~120dB, ~0.0005% THD. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Arny Krueger wrote:

> "Radium" <glucegen@excite.com> wrote in message > news:464c821f.0311091345.2c2c1f12@posting.google.com > >>What is the theoretically highest possible: >> >>1. Frequency Response (range of frequencies) > > > unlimited. You got the bits...
Unlimited frequency response just by adding bits? Surely not to the word length! One could argue that doubling the sample rate doubles the number of bits, but it doesn't seem you meant that. ... Jerry -- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;
If I am given a 16-bit fixed-point number of uniform quantization, I
can surely say that the maximum theoritic dynamic range I can get is
96.329 dB (assuming a voltage ratio and an constant impedance).  If I
am told the sampling rate is a uniform 8 KHz, I can say that the
maximum unaliased frequency I can derive is theoritically, less than 4
KHz.  Rather than say there is no therotical limit, let's just say we
don't have enough information to determine the limit.

Maurice Givens



Vladimir Vassilevsky <vlv@abvolt.com> wrote in message news:<3FAFACB2.F161A22F@abvolt.com>...
> Radium wrote: > > > > What is the theoretically highest possible: > > > > 1. Frequency Response (range of frequencies) > > > > 2. SNR > > > > 3. Dynamic Range > > > > 4. SPL > > > > 5. Musical Pitch (highest acceptable frequency) > > > > of PCM audio? > > There is no theoretical limit. > However there are different standards which specify the performance, and > the different hardware limitations. > > The today's state of art in the pro audio is ~24kHz, ~120dB, ~0.0005% > THD. > > Vladimir Vassilevsky > > DSP and Mixed Signal Design Consultant > > http://www.abvolt.com
Nope. If you are given even a 1-bit quantization, you can have any
dynamic range by multiple oversampling. If the sample rate is 8 kHz, you
can have multiple channels combined into one stream of any bandwidth.  
Vladimir Vassilevsky


DSP and Mixed Signal Design Consultant

http://www.abvolt.com



Maurice Givens wrote:
> > If I am given a 16-bit fixed-point number of uniform quantization, I > can surely say that the maximum theoritic dynamic range I can get is > 96.329 dB (assuming a voltage ratio and an constant impedance). If I > am told the sampling rate is a uniform 8 KHz, I can say that the > maximum unaliased frequency I can derive is theoritically, less than 4 > KHz. Rather than say there is no therotical limit, let's just say we > don't have enough information to determine the limit. > > Maurice Givens > > Vladimir Vassilevsky <vlv@abvolt.com> wrote in message news:<3FAFACB2.F161A22F@abvolt.com>... > > Radium wrote: > > > > > > What is the theoretically highest possible: > > > > > > 1. Frequency Response (range of frequencies) > > > > > > 2. SNR > > > > > > 3. Dynamic Range > > > > > > 4. SPL > > > > > > 5. Musical Pitch (highest acceptable frequency) > > > > > > of PCM audio? > > > > There is no theoretical limit. > > However there are different standards which specify the performance, and > > the different hardware limitations. > > > > The today's state of art in the pro audio is ~24kHz, ~120dB, ~0.0005% > > THD. > > > > Vladimir Vassilevsky > > > > DSP and Mixed Signal Design Consultant > > > > http://www.abvolt.com
"Vladimir Vassilevsky" <vlv@abvolt.com> wrote in message
news:3FB009D9.2C802828@abvolt.com...
> > Nope. If you are given even a 1-bit quantization, you can have any > dynamic range by multiple oversampling. If the sample rate is 8 kHz, you > can have multiple channels combined into one stream of any bandwidth. > Vladimir Vassilevsky > > > DSP and Mixed Signal Design Consultant > > http://www.abvolt.com >
wont the noise start adding up as well and ruin the SNR !
On Tue, 11 Nov 2003 04:06:02 GMT, malcolm wrote:
> > "Vladimir Vassilevsky" <vlv@abvolt.com> wrote in message > news:3FB009D9.2C802828@abvolt.com... >> >> Nope. If you are given even a 1-bit quantization, you can have any >> dynamic range by multiple oversampling. If the sample rate is 8 kHz, you >> can have multiple channels combined into one stream of any bandwidth. >> Vladimir Vassilevsky >> >> >> DSP and Mixed Signal Design Consultant >> >> http://www.abvolt.com >> > > wont the noise start adding up as well and ruin the SNR !
Not if you declare your 8kHz system to have infinite (or at least arbitrarily large) SNR (by making the number of bits per sample arbitrarily large). -- Andrew
On Tue, 11 Nov 2003 04:06:02 GMT, "malcolm"
<m.tamblynNOSPAM@mchsi.com> wrote:

> >"Vladimir Vassilevsky" <vlv@abvolt.com> wrote in message >news:3FB009D9.2C802828@abvolt.com... >> >> Nope. If you are given even a 1-bit quantization, you can have any >> dynamic range by multiple oversampling. If the sample rate is 8 kHz, you >> can have multiple channels combined into one stream of any bandwidth. >> Vladimir Vassilevsky >> >> >> DSP and Mixed Signal Design Consultant >> >> http://www.abvolt.com >> > >wont the noise start adding up as well and ruin the SNR !
Not if you shape it so that most of it is outside the audio band. That's how SACD works, and indeed it has very poor dynamic range above 20kHz. -- Stewart Pinkerton | Music is Art - Audio is Engineering
"Jerry Avins" <jya@ieee.org> wrote in message
news:boof7u$5q6$1@bob.news.rcn.net
> Arny Krueger wrote: > >> "Radium" <glucegen@excite.com> wrote in message >> news:464c821f.0311091345.2c2c1f12@posting.google.com >> >>> What is the theoretically highest possible: >>> >>> 1. Frequency Response (range of frequencies) >> >> >> unlimited. You got the bits... > > Unlimited frequency response just by adding bits? Surely not to the > word length! One could argue that doubling the sample rate doubles > the number of bits, but it doesn't seem you meant that.
It's exactly what I meant.