I promise, the quality difference is incredible …
Spanish version of this article in: https://oysiao.jlmirall.es/?p=4811
I have been researching intensively on how to improve the quality of online instrument lessons for a long time. I encourage you that if these posts helps you, invite me to a virtual coffee! At the end of this post you will find a button to help this site to remain free of ads and to finance the domain and hosting without being an extra cost for me.
The list of previous entries about online classes:
- La perspectiva de los alumnos en la docencia on-line musical
- Guía (no) definitiva para mejorar tu docencia on-line musical
- Desescalando en la clase de piano
- Exámenes de música en directo on-line. Instagram es una solución
- ¿Y si transformamos las clases on-line?
- Cómo hacer una lección a distancia para la enseñanza instrumental de música: Una Guía for Dummies por Lucia Di Cecca
- ¿Por qué se oye mal la música en las videoconferencias?
In the last post I presented an investigation that had been carried out on the audio quality of different platforms, in which Zoom was winning, but some teachers (such as the Valencian case) must use the official tools that our educational administration provides for us. And for video conferencing this platform is Webex. I have contacted Webex engineers directly to find solutions to the poor audio quality experience I had in my lessons, and they have not been able to help me so much. As they themselves admitted, all video conferencing software is designed for the spoken voice and therefore the rest of the frequencies and intensity differences are often lost along the way. Zoom has recently added an option to enable the original sound (so audio processing is minimal, and quality is better), but Webex doesn’t have a similar option (for now).
HOWEVER, I have found a solution that not even Webex engineers had even thought of!
This solution, requires two things:
- a computer for sending the audio (does not work with phones or tablets).
- a program (I explain it here with OBS) that bridges the microphone and Webex.
But before explaining the trick, some examples for you to believe me. As I said, the difference is brutal…
Original Sound 1: The sound that is intended to be sent in the first test:
Sound Received 1: The sound that has been streamed with normal conditions on Webex:
Original Sound 2: The sound that is intended to be sent in the second test:
Sound Received 2: The sound that has has been streamed with the new conditions in Webex:
Although the article analyzed in the previous posts had laboratory conditions (with equipment of more than 2,500 USD), in my case it has been a more “ecological” environment (in scientific research laboratory conditions are opposed to ecological ones, in the sense that the latter are those that occur outside the laboratories in natural environments).
I have used a Zoom H4N field recorder placed on top of the piano strings and faced to me. And I have made two video conferences by Webex, one with the usual conditions (the sound of the microphone is sent directly by Webex with the automatic volume adjustments disabled). And in the second video conference, the sound was sent through an intermediate application that was being shared on the screen and the Webex microphone was completely deactivated. In both cases, the sound sent by the Zoom H4n has been recorded on the source computer (indicated as original sound 1 and 2); and later on the destination computer a recording of the videoconference was made using Webex (indicated as received sound 1 and 2).
Let’s look at examples of both dynamics and frequencies.
You can clearly see how the sound is processed for streaming. Let’s look at the first Webex test with the usual sound settings:
In example 1, only spoken voice, we can already see one of the main differences. While the generated sound is Stereo (Left and Right), in Webex it is converted to Mono sound (a single channel).
At the same time, in these examples of spoken voice it can be seen (red box) how the volume peaks are more exaggerated compared to other phonemes that do not sound so loud and they remain with less energy as well.
In the second example, now piano in the low register. You can see how there is a fragment (red box) in which the sound disappears and when it returns it does so at a significantly lower volume. At the same time, at the beginning of the image (from the left), you can also see how the dynamics are less marked (that is, the proportions are not respected) and are more equal to each other.
Let’s see now, how these fragments improve in the second test with the new configuration that I have discovered.
In example 3, spoken voice again, we see how there is no difference in intensities. We also went from Stereo to Mono, but the changes between the original and the received sound are not as evident as before.
In Example 4, piano low notes, although there is a bit difference in intensities, no sound is lost anywhere. The dynamics change noticeably, but not excessively.
In the following images, it will be clear as how notes are disappearing in the usual configuration.
In this example of the low register, there is no doubt that there are times (red boxes) when the notes clearly go away (or just remains only the fundamental frequency and first harmonics).
And it is also appreciated how there is a sudden cut in frequencies from 8kHz. We recall that in the example of the paper in the previous post, in the Microsoft tools this cut was from 7kHz, and in the best possible setting of Zoom at 12.5kHz.
At the same time the signal / sound ratio is also worse in the received audio. This can be appreciated by looking at the “definition” with which the harmonics of the notes and the intermediate space can be visualized. In the original audio image the lines that indicate each of the harmonics are quite defined compared to those below that are more blurred. The more blurred, the more noise in the signal.
This example belongs to the end of the video conference where I perform an ascending arpeggio (with ups and downs) and also a crescendo. It can be clearly seen how the beginning (box on the left) is more blurred, there is a place where the bass directly disappear (center box) or even how the resonance of a chord is abruptly lost (box on the right).
Let’s now look at these same two examples with the new webex configuration.
If we compare what happens in Example 5 (the equivalent of this fragment) with Example 7, the difference speaks for itself. Although now the frequency loss occurs above 12kHz (and not above 8kHz), there are no notes disapearing.
And finally. Example 6 can be compared to Example 8. Although, with the normal configuration, the low frequencies disappeared and the resonance of a chord was cut off abruptly, now in Example 8 it no longer happens. If it is true that a little definition is also lost (it is enough to see how the lines of the fundamental notes of the arpeggio or even the resonance of the chord are more enhanced). In the box is where it is most clearly seen that in the received version there is a little more noise with respect to sound than in the original file, but not even close to the lousy levels of the configuration of example 6.
How to get this audio quality in Webex?
As has already been mentioned in numerous posts on this blog, all video conferencing programs are designed for the spoken voice, that is, they alter the sound they send so that a speaking person is better understood. But in the case of music, these modifications make life impossible for us. So you need to avoid making these modifications. Recently the Zoom platform has enabled an option called “use original audio”, which is what gives its advantage over the rest. But all teachers in the Valencian Community (Spain) must use Webex and for now, they do not have that option.
Directly not, but indirectly yes.
Another feature in which videoconferencing has improved in recent years is that of sharing the presenter’s screen. And if the presenter wants to share a video, … it must also be able to “capture” the audio of that video to retransmit it, and this is where all the excessive processes of audio transformation are NOT ACTIVE. If we are able to send our audio to the other end, not through the Webex general microphone but through another application and Webex captures the audio of that application, it is where we can get that increase in the quality of what we send.
In a moment I’ll explain a simple procedure to do it, but first I want to point out one thing. Whenever I have contacted the Webex engineers they have told me that nothing can be improved, because the codecs used on all platforms are to improve the voice, and that cannot be changed. THEY LIE! If they are able to retransmit sound with better quality when sharing a video from the computer, isn’t that a better quality audio transmission? So the excuse that it has been optimized for the voice no longer worth it. Zoom has seen it and that is why it offers that option. In Webex to be able to access that higher quality audio transmission we have to share the screen of our computer. The full screen or just the window of an application.
To achieve this I recommend using the OBS ap, for several reasons:
- It is a free and multiplatform application.
- It has been one of the great discoveries in software during this situation for many (including myself).
- If we have to share the screen of any application, it is most likely the OBS screen. Although it would also work by sharing the screen of another application, and with OBS sending the audio.
Install and open OBS.
The first thing you have to do is look at the controls below:
- Scenes: Creating a single scene is enough for what we want to do (right click, add scene).
- Origins: here we must indicate the micro that interests us. Right click in this area> add> capture the input audio> choose a name and accept> and on the device choose the name of the external microphone you are using (or the microphone of your computer if you have no other option).
- Audio mixer: In this part, increase the volume to 0.0 dB of the mic you have chosen in the previous block, and choose “advanced audio properties” by right-clicking. Again, look for the row corresponding to the mic you want to use and in “audio monitoring” select “monitoring and output”. In this way, if you also use OBS to record videos, your microphone will continue to be heard and you will not have to change the settings every time.
In this way, what your microphone captures will be heard by OBS, and when you capture the screen with Webex, that audio will be taken to send, but it will not carry out any transformation process (beyond a cutoff of frequencies above 12kHz and a little more noise than the clean signal of the micro, as we have already seen in the examples, but within what would be acceptable in terms of quality loss in a videoconference in 2020).
The same could be achieved with any application that is capable of making the microphone pick up live on your computer, but given the need to share the screen or window of your computer, it is very likely that you will have to share OBS.
If you want to use only one camera or be able to switch between several cameras (but not visible at the same time), you simply have to add each camera in origins, and create the scenes you need for each camera change. In each scene there can be a single origin.
In the View option from the top menu you can hide screen elements so that they are not retransmitted during the video conference, and thus leave only the main viewer.
Remember that it is convenient to make the videoconference with headphones (on both sides) to avoid echo and feedback problems.
And now how to configure Webex?
Once you have everything correct in OBS you can start Webex. Remember to deactivate your microphone (your voice will be heard by the microphone that we have just configured before through OBS, you do not need to select any audio source in Webex, since doing it from here we have all the known transformations problems for the spoken voice.
Start a meeting on Webex without audio and without image (the latter if you are going to use OBS for the image) and choose your headphones to listen.
Once inside the meeting, choose to share content:
Within the content sharing options, select these two options:
- Optimization for motion and video
- Share audio from your computer
As shown in the screenshot:
Then select the application window to share (OBS in the case that we explain here).
From this moment, the almost unprocessed audio signal will arrive on the other side from your microphone. If from the other side they also have to send you an audio signal with this quality, the exact same thing should be done.
Of course, there is a limitation, since access to this quality audio transmission must be done through the screen sharing function, we are subject to the limitations of this function. Two people cannot share the screen at the same time, with which an alternation of interventions must be managed to achieve this. One problem that may occur is that while you are not the one sharing the screen, you cannot be heard. I can imagine two ways to fix it:
- Brute force: At the moment when someone who is not sharing the screen, activates their option to share the screen, the one on the other end is disconnected and control is regained on your part.
- Being polite: Webex can deactivate the silence momentarily and that way we can be heard from the other side. So, politely ask for screen sharing.
I have been fighting with videoconferencing programs for a long time since last March, and I have always found the same excuse everywhere. “The transmission of audio data adjusts to the parameters of the voice and that cannot be solved.” Zoom has proven false, with its “allow original sound” option. And with Webex, if we use the audio sharing function of the computer, we also find an incredibly better transmission of audio data. Therefore: it is possible. It’s a tricky choice to make with the current interface, but that’s something that shouldn’t cost much to implement in future updates.
Video conferencing companies are still too focused on voice or Webinars, and do not understand that a transmission of audio data is necessary for other groups that goes beyond these limitations. They already have the necessary technology for this and it is latent in hidden functions of their applications. So it is totally false that it cannot be done. We need to get our demands across to solve these problems.
Call to action
So I thought that the best thing is to send this request directly to Cisco (Webex). I already have, and I encourage you to do so. The more people that do it, the sooner this problem will be solved, and those of us who are obliged to use this platform will be able to use it in appropriate conditions.
To forward your request to Cisco:
- Go to: https://mycase.cloudapps.cisco.com/attendeeCaseCreate
- Enter your e-mail so they can contact you with the answer.
- Choose: Webex Meetings and your Webex meeting URL. The Webex meeting URL is in your Webex App (Preferences>Account>URL)
- In “problem description” you can introduce this proposal that I leave below.
Hello, I am contacting you to request an improvement in the codec that is used by default to send the audio signal by videoconference. I am a music teacher and it is very difficult for me to teach with Webex. Competitive platforms such as Zoom allow the raw audio signal to be sent (this article demonstrates this: https://www.ianhowellcountertenor.com/preliminary-report-testing-video-conferencing-platforms) achieving much better quality and also in Stereo. And your app is also capable of doing this when the computer’s audio is shared by screen sharing (as demonstrated here: https://beyondthepiano.jlmirall.es/?p=296). It would be necessary to be able to access to this audio function from the main Webex window and be able to achieve higher audio quality without having to share the screen.
Hoping that you seriously consider this option to improve your product and keep it at the current quality standard in the market.
- In technology: Cloud and Hybrid Products > Webex (Meetings, Training, Events, Support)
- In Problem Area: Configuration > Software Failure
- Your name and phone
- Choose email as a form of contact
- And if you want, add a copy in “CC Recipients” so that the answer will also reach me and so I can also keep track of the times this request is made: firstname.lastname@example.org
- Show that you are not a robot in the captcha, and I think that no more field is needed to send the request