Subject: | |
From: | |
Reply To: | |
Date: | Sat, 29 Jan 2011 12:19:41 -0500 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
I'm always amazed that in the rush to embrace digital communication the
biggest "dirty little secret" of digital audio is forgotten: It only works
in a near perfect world. Analog audio can withstand all sorts of signal
degradation from selective fading to weak signals to interference, but it
doesn't take much for a digital signal to dissolve into total
unintelligibility.
Steve
----- Original Message -----
From: "Martin McCormick" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Saturday, January 29, 2011 09:22
Subject: Re: Anyone used d star
> The audio quality with digital voice communications
> depends on a ton of other factors. Straight PCM digital is what
> you have on CD's and movie sound tracks. We all know that that
> can be as good as it gets. Digital cell phones and two-way
> radios including DSTAR start out by digitizing your voice in
> straight PCM and then they get it ready for the limited
> bandwidth of sending it over the air. There are lots of schemes
> for compression and all of them loose sound quality, sometimes,
> lots of sound quality. One scheme I do know a little bit about
> is called LPC or Linear Predictive Coding. On the transmit end,
> your voice is analyzed and turned in to numbers representing
> frequency, amplitude and wave form information. At the receive
> end, a synthesizer turns those numbers back in to audio at the
> cost of sound quality due to the inevitable mathematical issues
> of rounding and timing that are part of anything this complex.
>
> Sure, you could send the straight PCM over the air but
> every voice channel would be about 100 KHZ wide or more. LPC
> encoding actually uses about 100 bits per second during silence
> and may go up to 9600 bits during sound which is much, much
> better as far as bandwidth.
>
> Also, you probably remember that the minimum band width
> for the human voice is 300 to 3000 HZ. That is what the
> telephone industry designs all their circuits to handle. You can
> use 8,000 samples per second for voice and it is pretty decent.
> If some system like LPC is built around that frequency range,
> the mathematical compromises plus occasional damage to the
> signal due to radio dropouts will make the sound pretty bad.
> Some of those digital systems fail interestingly in that a voice
> will suddenly turn in to weirdly musical sound like R2D2 from
> "Star Wars" because the receiving filter only got part of the
> information it should have gotten so it sends out the last good
> sample it got and sort of hangs. It is the audio equivalent of
> the picture freezes we get on digital television systems.
>
> One other thing to remember. You've also got to keep Mr.
> Nyquist happy. I'll let you ponder that last statement.
>
> 73
>
> Steve Dresser writes:
>> > And don't forget, your ts2000 is all digital as well for the audio
>> output,
>> > so don't be too quick to discount digital audio.
>>
>> I know that the noise reduction is digital processing in the audio
>> circuit,
>> and the high and low cut filters are digital as well, but I'm not sure
>> about
>> the rest of the audio circuitry.
>>
>> Steve
>>
>>
>
|
|
|