Many people remember that the series "Silicon Valley" tells about the programmer Richard
Hendrix, who accidentally came up with a revolutionary data compression algorithm and decided to
build your startup.
The show's consultants even came up with a metric by which to evaluate
such algorithms are the fictional Weissman Score.
Further down the story, the startup made a video chat using this solution.
The respected community is invited to discuss another, completely unusual
principle of data compression for audio and video calls, which solves the problem with the new,
unexpected side.
If you want to participate in the discussion of this decision, as well as find out what this
concepts with Jonathan Swift and the works of Leo Tolstoy, please under cat.
Some theory
Let's describe in general terms how modern audio communication works - the principle is the same as for
calls over GSM networks, as well as for instant messengers and VOIP networks.
Sound vibrations are sent to the microphone of the smartphone, then to the analog-to-digital
converter (ADC or ADC):
Next, encoding takes place with various codecs (G711, G729, OPUS, GSM, etc.),
encryption is added or not added (SRTP, ZPTP, etc.) and sent to the environment
data transmission.
For example, almost all messengers (WhatsApp, Viber, etc.) use the same codecs (recently, this is usually Opus), and almost the same slightly
modified protocols (based on SIP, WebRTC).
Both the public Internet and the GSM network can act as a data transmission network or
intranet:
Encryption is an optional element in this scheme, for example, in most cases for
SIP telephony does not use encryption.
But in messengers, on the contrary, they usually use their proprietary
protocols for encrypting voice and video.
Then the reverse process occurs - the addressee, having received the data, decodes the received information, then the signal goes to the DAC (digital-to-analog converter) and then goes to the audio amplifier connected to the speaker:
Characteristics of modern codecs:
G.711 64 Kbps
G.726 16, 24, 32 or 40 Kbps
G.729A 8 Kbps
GSM 13 Kb/s
iLBC 13.3 Kbps (30ms frame); 15.2 Kb/s (20ms frame)
Speex Range from 2.15 to 22.4 Kbps.
G.722 64 Kbps
Thus, for example, with a 7-minute conversation on WhatsApp or Skype,
consumed about 1 MB.
Let's remember these figures - 1Mb for 7 minutes of conversation, we will need them soon.
“Leo Tolstoy is like a mirror… of the revolution…”
Let's remember the most famous novel of this great Russian writer:
"War and Peace" - epic novel by Leo Tolstoy, describing the Russian
society in the era of the wars against Napoleon in 1805-1812. The epilogue of the novel brings
narrative before 1820.
The novel "War and Peace" by L.N. Tolstoy devoted seven years of intense and hard work. Manuscripts testify to how one of the world's largest creations was created.
"War and Peace": over 5200 finely written sheets have been preserved in the writer's archive.
If you want to read this novel now, you can easily download it.
And this file weighs only ... 1 MB:
The fb2 and epub formats, just like zip, rar, in principle, can be considered as a kind of
codecs.
Let's think about it - 7 minutes of our WhatsApp conversation is equal in terms of traffic
great work that took 7 years to write!
A conversation of 7 minutes was encoded with the opus codec, the novel was encoded with ePub, the volume is the same -
1MB, but what a huge difference!
Gulliver's Travels
Everyone knows this work by Jonathan Swift since childhood, but in fact this book is not for
children.
“Gulliver's Travels” is a political satire for adults, of course in the context of 18
century.
Surprisingly, Swift, being an ardent opponent of his other contemporary -
Newton, in his "Gulliver's Travels" not only predicted the discovery of satellites
Mars (with a fairly accurate description of their characteristics), but also described a rather interesting
way of communication between people:
“... the project required the complete abolition of all words;
the author of this project referred mainly to its health benefits and saving
time.
After all, it is obvious that every word we utter is associated with some wear and tear.
lungs and consequently leads to shortening of our lives.
And since words are only the names of things, the author of the project suggests that
that it will be much more convenient for us to carry with us the things necessary for expressing our
thoughts and desires.
… many very learned and wise people use this new way of expressing their
thoughts with things.
Its only inconvenience is the fact that, if necessary,
conduct a lengthy conversation on a variety of topics, the interlocutors have to carry on
shoulders large bundles with things, if funds do not allow hiring one or
two hefty guys. I have often seen two such wise men languishing under
the weight of the burden, like our peddlers. When they met on the street, they filmed
shoulder bags, opened them and, taking out the necessary things from there, conducted a conversation in this way in
the continuation of the hour; then they stacked their utensils, helped each other to load the load on
shoulders, said goodbye and dispersed.
However, for short and simple conversations, you can carry everything you need in your pocket.
or under the arm, and a conversation taking place at home does not cause any
difficulties. Therefore, the rooms where people who apply this method gather are filled with
all kinds of objects suitable to serve as material for such artificial
conversations.
Another great advantage of this invention is that it can be used
as a universal language understandable to all civilized nations, for furniture and household
utensils are everywhere the same or very similar, so that their use can be easily understood.
Thus, messengers can easily speak to foreign kings or
ministers whose language is completely unknown to them...”
So, you probably already guess what I'm talking about 🙂
Why transmit air tremors (sounds) for many hundreds and thousands of kilometers,
bother with encoding (in order to convey these air tremors to the addressee as accurately and efficiently as possible), keep the necessary bandwidth, if the semantic
Is the load of this transmission minimal, or even tends to zero?
After all, people communicate with each other not with sounds, but with meaning, content, semantics, thoughts…
The concept of the new communication system is quite simple - on the side of the source A, sound
fluctuations are also digitized, but are not immediately transmitted to the other side, but
are converted to text (Speech To Text) and then the already meaningful text is transmitted from
subscriber A, who:
- can be transmitted with the minimum required data transmission bandwidth (even HF type radio communication is possible, etc.)
- can be encrypted with any strong encryption algorithm
On side B, the received messages are decrypted and played back as a voice from
subscriber A (Text To Speech).
You can also download on side B so-called. subscriber A's voice avatar, which would
accurately repeated the manner of speech of subscriber A.
A separate channel can transmit background noises and emotions.
All the same is true for video communication - especially since individual elements have long been
exist in applications (various masks, backgrounds in Zoom, etc.).
Yes, there are technical aspects that are not yet fully implemented in the proper form -
for example, the speed of Speech To Text conversion will be critical, but using
predictive AI conversion algorithms can significantly increase this speed.
The most important advantage is the minimum bandwidth required in the transmission medium
data.
Those. this principle can be used not only for ordinary everyday
communications, but also for the military and for long-distance communications with long delays
(space communication, interplanetary - the Moon, Mars, etc. 🙂)
Although this is a description of the concept, in fact, in one of our projects there are already several
months, a prototype with this principle has been used.
But more on that next time...
Source: habr.com