1. What is the difference between speech signal attained from mobile
and that of telephone?

2. How to eliminate such terrible distinct difference? Is it possible?
Here I only wanna know the principle of algorithm ,for it can not be
fully explained in a post. And if you would like ,you can tell more.

3. Which institutions/Masters have been engaged in that study? The
present achievement?

Any comments would be appreciated!

Cheers
H...@gmail.com

HyeeWang wrote:

> 1. What is the difference between speech signal attained from mobile
> and that of telephone?

64kbit/s versus ~10kbit/s

> 2. How to eliminate such terrible distinct difference? Is it possible?

Yes. Just reduce the cellphone subscriber base to ~1/4 and increase the 
bit rate and the QOS accordingly. Cellular companies will love this 
initiative.

> Here I only wanna know the principle of algorithm ,for it can not be
> fully explained in a post. And if you would like ,you can tell more.

Google

> 3. Which institutions/Masters have been engaged in that study? The
> present achievement?

ITU, ETSI, TIA/EIA, ANSI.

> Any comments would be appreciated!

How much is the "would be appreciation" ?

VLV

On Apr 24, 10:12 am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> HyeeWang wrote:
> > 1. What is the difference between speech signal attained from mobile
> > and that of telephone?
>
> 64kbit/s versus ~10kbit/s
>
> > 2. How to eliminate such terrible distinct difference? Is it possible?
>
> Yes. Just reduce the cellphone subscriber base to ~1/4 and increase the
> bit rate and the QOS accordingly. Cellular companies will love this
> initiative.
>
> > Here I only wanna know the principle of algorithm ,for it can not be
> > fully explained in a post. And if you would like ,you can tell more.
>
> Google
>
> > 3. Which institutions/Masters have been engaged in that study? The
> > present achievement?
>
> ITU, ETSI, TIA/EIA, ANSI.
>
> > Any comments would be appreciated!
>
> How much is the "would be appreciation" ?
>
> VLV


Anyway, I am talking the speech quality! Not the communication
mode,not the bitstream rate.

Thank you Anyway!

Cheers

HyeeWang <h...@gmail.com> wrote:
 
> Anyway, I am talking the speech quality! Not the communication
> mode,not the bitstream rate.

In this newsgroup, you are supposed to be able to connect
bit rate and speech quality.

-- glen

>
>
>HyeeWang wrote:
>
>> 1. What is the difference between speech signal attained from mobile
>> and that of telephone?
>
>64kbit/s versus ~10kbit/s

The 64k speech quality on the wired network is poor. At less than half
that you can have pretty respectable wideband voice that makes land lines
sound bloody awful by comparison. AMR-WB on a 3G phone for example. Sadly
most cellular operators inhibit wideband now. In the early days of UMTS,
our local operators had it enabled to create some buzz around the new
networks that had masses of spare capacity. 3G to 3G calls sounded great.

>> 2. How to eliminate such terrible distinct difference? Is it possible?
>
>Yes. Just reduce the cellphone subscriber base to ~1/4 and increase the 
>bit rate and the QOS accordingly. Cellular companies will love this 
>initiative.

Most complaints about cellular voice quality relate to the bit error rate.
Current codecs operating from a clean bit stream aren't that bad. In some
parts of the world we seldom have significant error rates. In other people
hardly ever have a clean signal. Depends how stingy the network designers
are.

>> Here I only wanna know the principle of algorithm ,for it can not be
>> fully explained in a post. And if you would like ,you can tell more.
>
>Google
>
>> 3. Which institutions/Masters have been engaged in that study? The
>> present achievement?
>
>ITU, ETSI, TIA/EIA, ANSI.
>
>> Any comments would be appreciated!
>
>How much is the "would be appreciation" ?

Steve

Ha ha, you're kidding, right?  It's not all about the vocoder...

steveu wrote:
>>
>> HyeeWang wrote:
>>
>>> 1. What is the difference between speech signal attained from mobile
>>> and that of telephone?
>> 64kbit/s versus ~10kbit/s
> 
> The 64k speech quality on the wired network is poor. At less than half
> that you can have pretty respectable wideband voice that makes land lines
> sound bloody awful by comparison. AMR-WB on a 3G phone for example. Sadly
> most cellular operators inhibit wideband now. In the early days of UMTS,
> our local operators had it enabled to create some buzz around the new
> networks that had masses of spare capacity. 3G to 3G calls sounded great.
> 
>>> 2. How to eliminate such terrible distinct difference? Is it possible?
>> Yes. Just reduce the cellphone subscriber base to ~1/4 and increase the 
>> bit rate and the QOS accordingly. Cellular companies will love this 
>> initiative.
> 
> Most complaints about cellular voice quality relate to the bit error rate.
> Current codecs operating from a clean bit stream aren't that bad. In some
> parts of the world we seldom have significant error rates. In other people
> hardly ever have a clean signal. Depends how stingy the network designers
> are.
> 
>>> Here I only wanna know the principle of algorithm ,for it can not be
>>> fully explained in a post. And if you would like ,you can tell more.
>> Google
>>
>>> 3. Which institutions/Masters have been engaged in that study? The
>>> present achievement?
>> ITU, ETSI, TIA/EIA, ANSI.
>>
>>> Any comments would be appreciated!
>> How much is the "would be appreciation" ?
> 
> Steve
> 

steveu wrote:
>>
>>HyeeWang wrote:
>>
>>
>>>1. What is the difference between speech signal attained from mobile
>>>and that of telephone?
>>
>>64kbit/s versus ~10kbit/s
> 
> 
> The 64k speech quality on the wired network is poor. At less than half
> that you can have pretty respectable wideband voice that makes land lines
> sound bloody awful by comparison. AMR-WB on a 3G phone for example.

G.703 is definitely not the best use for 64kbit/s, however there is 
still the big difference in the throughput. Theoretically, a better 
coding scheme can be used.
The modern vocoders can work very well, this is not a question. However 
they work as expected only if the bit rate is sufficiently high, the 
error rate, the packet drop rate and the network latency are 
sufficiently low, and the incoming speech signal is clean.


> Sadly
> most cellular operators inhibit wideband now.

Their goal is to shove the maximum number of subscribers into the 
minimum of the infrastructure.

> In the early days of UMTS,
> our local operators had it enabled to create some buzz around the new
> networks that had masses of spare capacity. 3G to 3G calls sounded great.

AFAIR the improved voice quality was the strong argument for IS-95. 
Their codec at 14.4kbit/s sounded indeed much better then IS-54 VSELP.

> 
>>>2. How to eliminate such terrible distinct difference? Is it possible?
>>
>>Yes. Just reduce the cellphone subscriber base to ~1/4 and increase the 
>>bit rate and the QOS accordingly. Cellular companies will love this 
>>initiative.
> 
> 
> Most complaints about cellular voice quality relate to the bit error rate.

This also boils down to the density of subscribers per infrastructure.

> Current codecs operating from a clean bit stream aren't that bad. In some
> parts of the world we seldom have significant error rates. In other people
> hardly ever have a clean signal. Depends how stingy the network designers
> are.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

On Apr 24, 10:07 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
>
> AFAIR the improved voice quality was the strong argument for IS-95.
> Their codec at 14.4kbit/s sounded indeed much better then IS-54 VSELP.
>
IS-54's VSELP was a very early version of VSELP, snapshotted while
development was still in progress. A couple of years later it had been
massively improved. So much so, that they slowed the 8kbps rate down
to 5.95k for GSM half rate and to 4.2k for iDEN where it still sounded
better than IS-54.

That lack of fraction pitch estimation made the IS-54 codec sound
awful on some women's voices.

Steve

s...@coppice.org wrote:
> On Apr 24, 10:07 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
> wrote:
> 
>>AFAIR the improved voice quality was the strong argument for IS-95.
>>Their codec at 14.4kbit/s sounded indeed much better then IS-54 VSELP.
>>
> 
> IS-54's VSELP was a very early version of VSELP, snapshotted while
> development was still in progress. A couple of years later it had been
> massively improved. So much so, that they slowed the 8kbps rate down
> to 5.95k for GSM half rate and to 4.2k for iDEN where it still sounded
> better than IS-54.

Those two decisions (VSELP for HR and IDEN) look dubious from the 
technical standpoint. The ACELP codec would make perfect sense for GSM 
HR. For IDEN, a communication quality codec with the lower bit rate 
(like MBE or MELP) would be appropriate. Perhaps the decisions were made 
because of the patent wars.

> That lack of fraction pitch estimation made the IS-54 codec sound
> awful on some women's voices.

IS-54 codec is absurd.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

On Apr 25, 1:15 am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> ste...@coppice.org wrote:
> > On Apr 24, 10:07 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
> > wrote:
>
> >>AFAIR the improved voice quality was the strong argument for IS-95.
> >>Their codec at 14.4kbit/s sounded indeed much better then IS-54 VSELP.
>
> > IS-54's VSELP was a very early version of VSELP, snapshotted while
> > development was still in progress. A couple of years later it had been
> > massively improved. So much so, that they slowed the 8kbps rate down
> > to 5.95k for GSM half rate and to 4.2k for iDEN where it still sounded
> > better than IS-54.
>
> Those two decisions (VSELP for HR and IDEN) look dubious from the
> technical standpoint. The ACELP codec would make perfect sense for GSM
> HR. For IDEN, a communication quality codec with the lower bit rate
> (like MBE or MELP) would be appropriate. Perhaps the decisions were made
> because of the patent wars.

MELP was still in development at that time, and it was IMBE won the
public services applications (police, fire, etc) at low bit rate.
Having lost out on that, Motorola redirected their VSELP work for
MIRS, which became iDEN. The lack of patent royalties for them was
definitely a factor in that choice. For GSM HR. the VSELP codec won
fair and square against a number of CELP based competitors. ACELP was
not performing well at that time. People haven't worked out how to
tune the basic idea well. The lower bit rate option for G.723.1 was
settled about that time. Its ACELP based, and it sucks.

> > That lack of fractional pitch estimation made the IS-54 codec sound
> > awful on some women's voices.
>
> IS-54 codec is absurd.

Half finished is a better description. I wasn't involved at that time,
but I understand it was all done in a rush, and nobody else was in a
better position at that moment to beat it in the runoff. I guess its
competitors were even more absurd. :-)

Regards,
Steve