Q: Op-amp for high quality voice ??

mas1738 said:

As far as i know that the human voice has a frequency up to 3.4kHz however the human can hear up to 18-20kHz (for example: music)

Click to expand...

No.
All the important consonants sounds in speech are well above 3.4kHz:
s, sh, t, th, p, f, h, j, k, v, etc.

In Google there are some funny articles about what people are guessing about what was said when the bandwidth is too low.

audioguru said:

No.
All the important consonants sounds in speech are well above 3.4kHz:
s, sh, t, th, p, f, h, j, k, v, etc.

Click to expand...

You mean all those have some content above 3.4KHz, almost all of it though is well below that.

Bandwidth limiting for voice communication is normal, and provides perfect communication quality - if it wasn't for bandwidth limited audio we wouldn't have telephones, either home or mobile.

mas1738 said:

So you advice me not to digitize the signal, because the project will be marked by measuring the SNR at the receiver so if the analog signal will give me better SNR I will go for it, since I didn’t build the circuit yet

Click to expand...

If you're only digitising with 6 bits the quality is going to be absolutely appalling, really heavily distorted. If you want to use digital sound then use at least 8 bits, and preferably more.

Bear in mind though, using digital will greatly increase the required bandwidth of your transmission path, and the higher the bandwidth the shorter the range.

I can't remember who said that speech has important frequencies to 14kHz.

Meyer Sound says to 10kHz:

EDIT:
Polycom, the speakerphone people say that consonants in speech go up to 14kHz and are very important for intelligibility.
https://www.electro-tech-online.com..._of_bandwidth_on_speech_intelligibility_2.pdf

Nigel Goodwin said:

If you're only digitising with 6 bits the quality is going to be absolutely appalling, really heavily distorted. If you want to use digital sound then use at least 8 bits, and preferably more.

Click to expand...

Is the 6/8 bits you are talking about equal the "bit-depth", sample rate or both?

From what I understand, the formula for calculating Bits Per Second is

BPS = (Bit depth x sample rate x number of channels.)

Or am i way off here?

Using a dedicated comparator would be best, mabye the LM311 advertised as a "high speed" device. Take care with the pinout though! Otherwise I'm sure a TL071 could be used.

Out of interest I used a VST "quantiser" effect to simulate 6 bit quality on a voice recording. The effect was audible but not massive. This is a high quality normalised level voice recording though. I also tried filters at 3.6k as well as setting the quantiser to 6khz sampling rate. Both of these were found to be more detrimental to intelligibility than the 6 bit.

My advice would be to use a good quality AGC (automatic gain control) circuit if you are transmitting with just 6 bits. This should make the system quite acceptable. Also don't use and low pass filters and try to keep the sampling rate high (if that is relevant to this application i'm not sure)

Hello again too! Haven't visited in a while. Nice to see audioguru and Nigel still here

audioguru said:

I can't remember who said that speech has important frequencies to 14kHz.

Click to expand...

What do you mean by 'important'?.

Perfectly intelligible speech requires only 3 to 4KHz as already mentioned, increasing that simply limits the range and the space available for multiple users.

I would like to think that comunication to aircraft is perfectly clear, and they use crappy low bandwidth AM radio

Those fricatives are mostly above 3.4khz. You lose sibilance when you cut it down like that.

gabeNC said:

Is the 6/8 bits you are talking about equal the "bit-depth", sample rate or both?

Click to expand...

I'm presuming that by 6 bits he means the resolution of the A2D conversion, which only gives a possible 64 levels - so very poor quality - sample rate has nothing to do with that.