• Home
  • Tablet
  • Ultra-high-quality Internet video distribution Live Extreme. What is the idea of ​​“delivering as it is”?
24 Mar, 2023

Ultra-high-quality Internet video distribution Live Extreme. What is the idea of ​​“delivering as it is”?

Live Extreme Accelerates

--Since our last interview in July, the number of platforms seems to be increasing.

Mr. Oishi (honorific titles omitted): Well, in addition to IIJ, which we have been working with for a long time, I think we were talking about starting a service with Thumva, but we will be coordinating with Teruhi's eContent in September. Announced. In fact, we had been testing the connection with the server since around spring, but we started it when the service system was ready. Shortly after that, we also announced a partnership with J Stream. Like IIJ, J-Stream is a major player supporting the bottom line, and has a platform that can withstand viewing by thousands or even tens of thousands of people.

In addition to simply distributing, we provide a one-stop service for everything from ticketing to event implementation, online distribution, creating a customer database, analyzing customer information, conducting questionnaires, sending DMs and selling goods. We also have a service called Ibekas, and Live Extreme has entered as one of the menus, so it seems that various developments will be possible in the future.

－－On the other hand, you also do various interesting streams, right?

Oishi: After Emi Fujita's paid streaming, we are streaming the first live show of The Cosmic Suite Ensemble led by Yasushi Ide from Thumva. Furthermore, in November, Ryo Konishi's solo project, Zominsha, was live-streamed.

Here, for the first time, we used HPL to distribute the main audio binaurally, and for the first time, we also distributed the secondary audio. As soon as the live distribution ended, Mr. Konishi entered the studio and watched his own distribution video, commenting on the songs and telling secret stories, making it a very interesting content for the fans.

It is said that in the case of live music distribution, 70% of the tickets for real-time listening are sold, and 30% are sold during the archive period. A phenomenon occurred in which “double the real time” was sold. I believe that this experiment will serve as a guideline for the future live music distribution business.

Also, November 11th was “Bass Day”, so we did a free distribution related to the bass. This is the first time in a long time that we have hosted a distribution, and we use eContent. Specifically, we prepared two types of distribution formats: full HD + high resolution (96kHz/24bit) and full HD + lossless (48kHz/24bit), and distributed to a limited number of 1,000 people. And on December 4th (*note, the interview was on December 2nd), we will be streaming the alternative band Hitsuji Bungaku. This is the second version using eContent, but this time I'm going to try something interesting.

Sheep Literature Live

--- What do you mean by "interesting experiment"?

Oishi: This time, we will be performing at Billboard Live Yokohama, and Billboard Live and LIVE LOVERS will form a tag team. Here, in addition to normal distribution with Streaming+, premium high-definition and high-quality sound distribution using Live Extreme will be performed. Regular distribution is 2,800 yen (2,400 yen for Club BBL members), and premium is 3,200 yen (2,800 yen for Club BBL members), a difference of 400 yen. I'm thinking of doing some sort of marketing survey to see how much difference there will be in sales.

Hitsuji Bungaku distribution site on December 4th

--How much difference is there between normal distribution and premium distribution?

Oishi: Actually, the origin is the same, the video is full HD through SDI, and the audio is mixed/mastered at 96kHz/24bit, but normal distribution = Streaming+ via analog, premium distribution = Live Extreme Receive digital via AES/EBU.

This is encoded by each and launched separately = uploaded. While Streaming+ uses H.264 for video and AAC for audio, our Live Extreme uses H.264 for video and lossless audio for a total of 8Mbps. Since this 8 Mbps is delivered to the viewer as it is, there is a considerable difference in both sound quality and image quality.

－－Since the audio is AAC and lossless, I understand that there will be some differences, but the video is the same for both normal and premium distribution because it is full HD H.264, right?

Ultra-high-quality Internet video distribution Live Extreme What is the idea of ​​``delivery as it is''?</p><p>Oishi: If you actually compare them, you'll understand, but this is quite different. There may not be a big difference until the H.264 encoding and launching, but in normal distribution, the server side re-converts, which inevitably results in degradation. This is the same for audio as well as video, and the difference between lossless and lossless is quite large because what is converted to AAC on site is then reconverted on the server.</p><p>－－It seems that re-conversion on the server is very wasteful and has a negative impact on image quality and sound quality, but what does that mean?</p><p>Oishi: Mr. Fujimoto also broadcasts on DTM Station Plus!, so I'm sure you're familiar with it. This is a protocol originally developed for video distribution to Macromedia's (currently Adobe) FLASH player. Now that FLASH is gone, RTMP cannot be played back, so streaming servers convert it so that it can be viewed in a browser.</p><p>Specifically, real-time conversion to formats such as HLS (HTTP Live Streaming) and MPEG-DASH (Dynamic Adaptive Streaming over HTTP) is delivered to the client, and the contents are H.264 and AAC. Common. If it originally came in H.264 or AAC, reconverting it may seem like a waste, but there's a big reason for that. One of the reasons is Adaptive Bitrate. Adaptive Bitrate is a technology that changes the amount of information (compression ratio) of content in real time according to the communication status on the playback client side.</p><p>－－When the line condition is bad, the streaming will not stop, but the image quality may suddenly drop.</p><p>Oishi: That's right. For example, if you create 3 types of 1Mbps, 4Mbps, and 8Mbps on a streaming server and stream them at the same time, you can switch them at any time according to the communication environment of the client, so that the image and sound do not stop and the optimum image and sound are produced. It will be possible to watch while switching to audio. In that case, there should be a way to pass through the original signal without re-converting it, but in the world of programming, making an exception would complicate the processing, so re-converting it is unavoidable. and result in deterioration.</p><p>While doing this job, I'm always curious about what's going on with various live videos, so I'm watching various concerts. Under such circumstances, it is the live performances of idols that stand out. There are many productions that use lasers, and in such cases, block noise is generated by encoding twice. On the other hand, even if the audio is AAC with the same bit rate, encoding twice will result in a large deterioration in sound quality, and the difference with lossless is particularly large.</p><p>--Is there a way to easily compare how much the sound quality actually drops?</p><p>Oishi: If you compare the normal distribution and premium distribution of Hitsuji Bungaku this time, I think you can feel the difference. On the other hand, I did a little experiment the other day. I wrote about it on my blog the other day, but the audio settings for the live encoder recommended by YouTube are as shown in the figure. When I live-streamed on YouTube with this recommended audio setting (AAC-LC 128kbps) and H.264 (1080p30) video, 6 types of data were generated by Adaptive Bitrate. The contents are shown in the figure below, but if you look closely, the video is in 6 stages, but the audio is HE-AAC (44.1kHz, 2ch, 48kbps) and AAC-LC (44.1kHz, 2ch, 128kbps). There are only types.</p>Six types of data were generated by the audio setting Adaptive Bitrate of the live encoder recommended by YouTube<p>--It's simpler than you think. However, according to what I just said, when you first launch it, it will be converted to so-called AAC, that is, AAC-LC, and then converted to high-compression HE-AAC or general AAC-LC on the media server. Isn't it?</p><p>Oishi: That's right. At that time, I tried to see how much the sound would differ, how much the waveform would change by inputting a square wave, and how the spectrum would change. This is the original square wave. Here is what I converted to AAC-LC 128kbps with my live encoder. Looking at this, you can see that the square wave is slightly disturbed and the spectrum is gradually decreasing above 17kHz.</p><p>I uploaded it and this time I reconverted it to AAC-LC on the streaming server. Although it is the same AAC-LC 128kbps, you can see that the square wave is further disturbed and above 16kHz is cut off. This is the sound quality deterioration due to re-conversion.</p><p>Furthermore, you can see that the conversion to HE-AAC 48kbps is considerably degraded. Here is a comparison of the AAC-LC conversion results on the live encoder and the conversion results to AAC-LC on the streaming server overlaid. I think you will find that it is quite different. On the other hand, in the case of Live Extreme, you can see how much difference it is because it is lossless and delivered to the end.</p>Converted to AAC-LC 128kbps with the original square wave live encoder Converted back to AAC-LC on the streaming server , and compared the results of conversion to AAC-LC on the streaming server by overlaying them<p>--This is a very easy-to-understand comparison. In the case of Live Extreme, we don't think about Adaptive Bitrate, and the quality of sound and image quality is good because we receive what we put in anyway.</p><p>Oishi: We are very particular about bit perfection, so if it gets re-encoded on the streaming server, we're out. That's why, while preventing this from happening, we directly encode HLS or MPEG-DASH format within the live encoder and upload it to the distribution server so that you can watch it directly in your browser without the need for special player software. With this, you can play it in the browser as it is, so the distribution server just delivers it without changing anything. That's why you can watch it without changing the image quality or sound quality.</p>Encode directly to HLS or MPEG-DASH format in the live encoder and upload to the distribution server For 48kbps, the deterioration is not worse than expected.</p><p>Oishi: HE-AAC maintains a certain amount of sound quality while keeping the bit rate small. Actually, in order to halve the sample rate, even if 48kHz is sent, it is internally processed at 24kHz. But when decoding, we return it to 48kHz and put in a system that complements the missing high frequencies. Of course, the high-frequency components will be different from the original, but when you try listening to the sound without compensation, it sounds pretty terrible, so it's a fairly high-precision compensation technology. However, it seems that the spectral envelope is preserved before dropping to 24kHz, so it seems that that part is well reproduced.</p><p>－－If there is an opportunity next time, I would like to experiment with the sound quality of HE-AAC myself. We look forward to seeing how Live Extreme develops in the future.</p><p>Oishi: I think HE-AAC is an interesting experiment. Live Extreme will be adopted for more live distribution in the future, and we plan to take various approaches to reach more people, so please look forward to it.</p></div> </p> <!-- Start Post Pagination --> <div class= Prev Next