All this voice stuff goes back to copper lines. Some random thoughts that I heard circa 1970 while working for Ma Bell and T-1 was the new kid on the block. When extending trunks the two wire line went through a hybrid that converted it to four wire. Two for transmit and two for receive. Amplifiers were used. Higher frequencies attenuate at a greater rate than lower frequencies. The signal was filtered into bands with each band amplified at a different gain so that at the receiving end it sounded normal. Choices had to be made about how high a frequency to use. A copper pair has resistance (length) and capacitance (each pair is twisted together to minimize crosstalk.) To offset the effects of capacitance load coils (inductance) was added. Thus an RCL tuned circuit. It was optimized for frequencies below 4khz. This work predates digital carrier. Side note: Analog frequency seperated carriers used inband signalling. Various devices had different functions but the common point was the use of a 2600 Hz tone. Some folks had too much of that frequency in their voice and so disconnected themselves.
> -----Original Message----- > From: David L. Blair [mailto:[EMAIL PROTECTED]] > Sent: Wednesday, February 27, 2002 9:58 AM > To: [EMAIL PROTECTED] > Subject: Re: Re: Sample Rate [7:36566] > > > > John Neiberger wrote: > > What I'm trying to find out is why the original 4KHz limit on > > voice calls was put into place. It sounds like it was simply > > an arbitrary decision. 4KHz is sufficient for a telephone call > > and to provide clear calls that included higher frequencies > > might have added some technical complexities, perhaps. > > > > They also added a high-pass filter around 400Hz since most > > telephones can't reproduce low frequencies well and it also > > filters out some harmonics of 50-60Hz hum that might show up > > from time to time. That is concrete reason for including a > > high-pass filter and I wondered if there was a concrete > > technical reason for including the 4KHz low-pass filter. From > > the sounds of it there really isn't a technical issue, 4K is > > just a nice round number. :-) > > I used three sources to answer John's query: "Voice over IP > Fundamentals", > "Cisco Voice over Frame Relay, ATM, and IP", and Integrating > Voice and Data > Networks". These are great books for anyone wanting to know > more about > voice technologies. > > Interesting Facts and Ideas I came across: > > 1) Human hearing is in the range of 200 Hz to 20,000 Hz > > 2) Human speech is in the range of 250 Hz to 10,000 Hz. Most of the > information comes from the middle frequencies. According to > Nyquist, "Human > voice contains sounds that are more often Middle-pitched > frequencies than > either High or Low pitched frequencies. > > 3) Frequencies greater than 4,000 Hz are filter out to limit > crosstalk. > > 4) During the Analog to Digital conversion voice samples are > put though a > process called Quantization. Quantization is the process of rounding > sampled values to the nearest predefined discreet value. Pulse Code > Modulation (PCM) is a Quantization process. PCM is also used > to achieve 12 > to 13 bits of voice information in 8 bit words. Two commonly > used PCM's are: > mu-law (North America), and a-law (Europe). What you hear is > not someone's > voice, but a representation of their voice. > > 5) Noise is a major issue when talking about voice quality. Noise is > constant problem for Analog signals. What is signal and what > is Noise? > When a Analog signal is amplified so is the Noise, which in > turn makes the > quality of Analog calls worst as the distance increases. > Digital Calls are > less suitable to Noise than Analog calls. > > 6) Delay is a major issue when talking about conversation flow for two > reasons: 1) For a conversation to flow normally, the delay is > receiving the > voice information must be less than 250ms. When the delay is > more than > 250ms, the human receiving the voice message will start to > talk thinking the > human sending the voice message is at a breaking point in the > conversation, > i.e.. both people are talking at the same time similar to a > collision in > Ethernet. Delay is also important in how the voice packets are filled > during the Analog to Digital conversion. That is why ATM > (ATM cell is 53 > octets, 5 octets are header and 48 octets are payload) is a > good method for > transporting voice packets because the delay to fill the > payload section is > smaller than with other cell/packet types. > > > Answer: It does indeed seem that the 4,000 Hz mark was > arbitrary in nature; > 3,500 Hz or 5,000 Hz would work also. It is a "nice round" > number to work > with. Simplies any math work. Middle frequencies carry the > bulk of the > information and Human speech upper limit is 10,000 Hz amd > 4,000 is near the > middle. The low filter is also to reduce the frequencies that > carry less > information. > > Hope this helps. > > > "Through Complexity there is Simplicity, > Through Simplicity there is Complexity" > > David L. Blair - CCNP, CCNA, MCSE, CBE, A+, 3Wizard Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=36667&t=36566 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
