NOTICE OF UNLICENSED OPERATION

[Lscott]

Don’t forget WiFi, Bluetooth and satellite radio/TV.

[tcp2525]

3 hours ago, Lscott said:

You can quit using your cellphone. It’s based on wireless two-way radio and digital voice technologies.

Oh trust me, I was pissing and moaning when they transitioned to digital. The sound quality of the analog was far superior to digital. Not a problem now, we're all programmed to accept crappy sounding phone calls. What choice do we really have?

[Lscott]

16 hours ago, tcp2525 said:

What choice do we really have?

Two cans connected by a string still works.

[Lscott]

Unfortunately just about everyone uses some form of a digital voice encoder. The leading favorite is the AMBE, previously IMBE, by DVSI. Their proprietary codec is based on early work done at MIT.

The link below is a short description from DVSI, which doesn't really revel much.

https://www.dvsinc.com/papers/iambe.html

A much more in-depth description can be found here from a report published by MIT for the US Air Force in 1987.

https://apps.dtic.mil/sti/tr/pdf/ADA181146.pdf

My understanding is when the FCC forced the commercial radio services to move to narrow band FM, which was done by reducing the FM deviation, also resulted in a reduction in the signal to noise ratio I believe. At a 12.5KHz bandwidth it's not severe, however at narrower bandwidths it is. The FCC stated at one point they intend to move to a true 6.25 KHz per voice channel width. That's why the major radio manufactures introduced various digital voice technologies. For the moment the FCC seems to be OK with various digital voice technologies that can achieve an "equivalent" voice channel width of 6.25 KHz, example DMR 2 slots in a 12.5 KHz channel. However at some point they may force a move to a true 6.25 KHz channel width, but no official date has been given.

This is stated in chapter 1 of the FCC's narrow banding guide.

https://transition.fcc.gov/pshs/docs/clearinghouse/guidelines/Narrowbanding_Booklet.pdf

In the mean time there are ways to license a true 6.25 KHz channel per the FCC. See attached paper.

 

 

Splitting 6.25KHz Channels.pdf

[tcp2525]

2 hours ago, Lscott said:

Two cans connected by a string still works.

And the audio quality is excellent. 

[tcp2525]

1 hour ago, Lscott said:

Unfortunately just about everyone uses some form of a digital voice encoder. The leading favorite is the AMBE, previously IMBE, by DVSI. Their proprietary codec is based on early work done at MIT.

The link below is a short description from DVSI, which doesn't really revel much.

https://www.dvsinc.com/papers/iambe.html

A much more in-depth description can be found here from a report published by MIT for the US Air Force in 1987.

https://apps.dtic.mil/sti/tr/pdf/ADA181146.pdf

My understanding is when the FCC forced the commercial radio services to move to narrow band FM, which was done by reducing the FM deviation, also resulted in a reduction in the signal to noise ratio I believe. At a 12.5KHz bandwidth it's not severe, however at narrower bandwidths it is. The FCC stated at one point they intend to move to a true 6.25 KHz per voice channel width. That's why the major radio manufactures introduced various digital voice technologies. For the moment the FCC seems to be OK with various digital voice technologies that can achieve an "equivalent" voice channel width of 6.25 KHz, example DMR 2 slots in a 12.5 KHz channel. However at some point they may force a move to a true 6.25 KHz channel width, but no official date has been given.

This is stated in chapter 1 of the FCC's narrow banding guide.

https://transition.fcc.gov/pshs/docs/clearinghouse/guidelines/Narrowbanding_Booklet.pdf

In the mean time there are ways to license a true 6.25 KHz channel per the FCC. See attached paper.

 

 

Splitting 6.25KHz Channels.pdf 103.57 kB · 0 downloads

Of course it makes sense to do it this way. After all it is just voice communications, not audiophile quality needed for land mobile service. I just hate the compression it causes. A lossless CODEC would help sound quality at the expense of bandwidth. 

 

[Lscott]

40 minutes ago, tcp2525 said:

Of course it makes sense to do it this way. After all it is just voice communications, not audiophile quality needed for land mobile service. I just hate the compression it causes. A lossless CODEC would help sound quality at the expense of bandwidth. 

 

I'm sure if it was possible that's what would have been done. Some of the comments I've read said the DVSI code was selected also due to it's low bandwidth requirements, a feature of the algorithm used. That allowed decent sound quality in the allowed bandwidth. At the time some felt it wasn't possible even using digital methods.

The simple version of DVSI's algorithm involved a quick analysis of a snapshot of a small time slice of the audio. The resulting info was used to derive variables that were then transmitted. Those variables then were inserted in an algorithm that "simulated" the human vocal track as a filter for multiple sine wave and noise sources. The output of the filter is a simulation of original human speech. That's why it sounds a bit weird because its NOT a compressed direct digital conversion of the original voice.

One of the main complaints is the simulated speech lacks some of the subtle nuances of the original speaker, thus for some people making it difficult to tell who it is they are hearing, even though the speech is very readable.

One other problem is the algorithm is highly optimize for human speech ONLY.  Back ground sounds, like wind noise, sirens etc confuse the crap out of the the process. I was reading a long thread on another forum years back where firefighters were VERY concerned about this. The radio manufactures had to implement various solutions in their radio's audio path to mitigate those issues. Some did a better job that others. This likely accounts for the comments where people claim some digital modes sound better even though they use the SAME EXACT codec. The sound quality likely even varies between manufactures using the same digital voice mode and codec.

Also don't forget that modes like DMR uses time slicing, i.e. TDMA slots, so the number of available bits that could be used for improved sound quality are missing verses a mode like P25 which has a higher bit rate I believe.

Each digital mode has a fairly complex signaling scheme for communications, which of course consumes bits which could be used for better sound quality by transmitting more parameters to be used in the voice reconstruction process. Some of those bits are used for error correction. Also when Motorola grafted encryption on to MotoTrbo they had to use some of the error correction bits for the encryption info. Some have noticed when DMR enhanced encryption is used the voice quality can degrade a bit.

One can really go down the rabbit hole on this topic. It's not as simple as it first appears.

[tcp2525]

1 hour ago, Lscott said:

I'm sure if it was possible that's what would have been done. Some of the comments I've read said the DVSI code was selected also due to it's low bandwidth requirements, a feature of the algorithm used. That allowed decent sound quality in the allowed bandwidth. At the time some felt it wasn't possible even using digital methods.

The simple version of DVSI's algorithm involved a quick analysis of a snapshot of a small time slice of the audio. The resulting info was used to derive variables that were then transmitted. Those variables then were inserted in an algorithm that "simulated" the human vocal track as a filter for multiple sine wave and noise sources. The output of the filter is a simulation of original human speech. That's why it sounds a bit weird because its NOT a compressed direct digital conversion of the original voice.

One of the main complaints is the simulated speech lacks some of the subtle nuances of the original speaker, thus for some people making it difficult to tell who it is they are hearing, even though the speech is very readable.

One other problem is the algorithm is highly optimize for human speech ONLY.  Back ground sounds, like wind noise, sirens etc confuse the crap out of the the process. I was reading a long thread on another forum years back where firefighters were VERY concerned about this. The radio manufactures had to implement various solutions in their radio's audio path to mitigate those issues. Some did a better job that others. This likely accounts for the comments where people claim some digital modes sound better even though they use the SAME EXACT codec. The sound quality likely even varies between manufactures using the same digital voice mode and codec.

Also don't forget that modes like DMR uses time slicing, i.e. TDMA slots, so the number of available bits that could be used for improved sound quality are missing verses a mode like P25 which has a higher bit rate I believe.

Each digital mode has a fairly complex signaling scheme for communications, which of course consumes bits which could be used for better sound quality by transmitting more parameters to be used in the voice reconstruction process. Some of those bits are used for error correction. Also when Motorola grafted encryption on to MotoTrbo they had to use some of the error correction bits for the encryption info. Some have noticed when DMR enhanced encryption is used the voice quality can degrade a bit.

One can really go down the rabbit hole on this topic. It's not as simple as it first appears.

It's definitely fascinating tech. I can see why digital is so appealing, but I just can't get used to the sound. 

[SteveShannon]

28 minutes ago, tcp2525 said:

It's definitely fascinating tech. I can see why digital is so appealing, but I just can't get used to the sound. 

Everyone sounds drunk.