Seeking Extensive and Specific Stream Information

[Corwyn Allen]

Greetings,

I am looking to solve several issues in SL vis a vis using streams to present stories read by multiple readers.

Some background first. I have a stream in SL and have been a performing musician for some time now. I have also been doing extensive reading and acting in SL using voice via the SL and Vivox servers. This latter, the Vivox server, is becoming an issue as it is increasingly breaking down. It is becoming less reliable for our storytelling groups. Next, I have a fairly extensive technical background in RL as a network engineer and a video conferencing engineer that required knowledge of studio grade audio. I have had a home studio set up for a number of years prior to getting a stream in SL where I did voice acting and recording of audio books. I have a small mixer and a professional studio mic and headset. 

Now, on to the things I need more expertise in. First of all, what are some ways to troubleshoot a stream issue in SL. I pretty much know how to troubleshoot my own equipment (PC, mixer, et. al.) but am at a loss as to how to determine if a drop of the stream is a streaming service issue, a parcel issue, or an individual listening user issue. What are some methods streamers use to determine the source of a problem and ultimately its fix?

Next, I have performed music with other musicians using stacked streams. This is not a viable option for readers as all the readers need to be able to hear each other, and in stacked streams only the last person in the stream, the one who is feeding their stream to the parcel, can hear everyone. Those farther up the stack can only hear those above them until the first person can only hear themselves and no one else. So I would like to know if anyone has been able to join multiple streams together so that everyone in the streamed group can hear everyone else. If so what are the lag issues you have dealt with. Have you used a third party software or service to accomplish this? If so, what sw or service? 

To end and clarify I need to be able to have two to ten readers on streams able to hear each other in order to read various parts together. Stacking is not viable. I also need to know what you're using to troubleshoot stream issues in SL. 

Thank you,
Corwyn Allen

[Oz Linden]

21 hours ago, Corwyn Allen said:

I have a stream in SL and have been a performing musician for some time now. I have also been doing extensive reading and acting in SL using voice via the SL and Vivox servers. This latter, the Vivox server, is becoming an issue as it is increasingly breaking down.

Our monitoring doesn't show that. If you are having repeated problems that you can describe more specifically, we could try to help diagnose them, but Jira is a better way to communicate about them than the forum.

If you have local recording capability on your system, recording the audio for one of your sessions (with the informed permission of the participants) would be one way to start diagnosing voice issues (many problems produce recognizable audio artifacts). We could also help you learn to do packet captures of the relevant network traffic for analysis.

[Corwyn Allen]

7 hours ago, Oz Linden said:

Our monitoring doesn't show that. If you are having repeated problems that you can describe more specifically, we could try to help diagnose them, but Jira is a better way to communicate about them than the forum.

If you have local recording capability on your system, recording the audio for one of your sessions (with the informed permission of the participants) would be one way to start diagnosing voice issues (many problems produce recognizable audio artifacts). We could also help you learn to do packet captures of the relevant network traffic for analysis.

Thanks Oz, but this topic isn't about fixing Vivox. It's about find more expansive uses of stream. 

[Callum Meriman]

On 10/12/2018 at 2:47 AM, Corwyn Allen said:

Next, I have performed music with other musicians using stacked streams. This is not a viable option for readers as all the readers need to be able to hear each other, and in stacked streams only the last person in the stream, the one who is feeding their stream to the parcel, can hear everyone. Those farther up the stack can only hear those above them until the first person can only hear themselves and no one else. So I would like to know if anyone has been able to join multiple streams together so that everyone in the streamed group can hear everyone else. If so what are the lag issues you have dealt with. Have you used a third party software or service to accomplish this? If so, what sw or service? 

Interesting problem! I can't speak from experience of doing it, although I stream a lot and can visualise some of the issues. I hope you don't mind me making these comments. Apologies in advance if it's useless.

Solution 1.... Streaming - lowcost, complicated, laggy, likely not ideal

You would need everyone to use similar software that would transmit their voice reliably. An Icecast endoint using Mutt for example.

A central person - the sound engineer - would attach to all the endpoints. It would be a star shape. And that central person would open up each and every performer's stream. He would rebroadcast his systems audio to a proper shoutcast server with sufficient capacity for the audience.

It would be ideal if he were using Linux and Jack so that each of the performers streams could be fed to an onscreen mixer, and his stream fed to the shoutcast server. I am not sure Windows is smart enough to do a good job of it.

There are two problems - One will be foldback, the performer can deal with this by removing their headphones as they speak, but that will be an issue with snappy dialog.

The other problem, and a far bigger one, will be lag, and I can't see a way around that in a live situation. There is a 1.5 second or so delay at each point, 6 seconds of lag all in all.

Performer ----- 1.5 seconds ------> technician -------- 1.5 seconds --------> other performers -------> 1.5 seconds ----> technician ---- 1.5 seconds -----> 1st performer

Solution 2... Workable, lag free, low quality voice only, but would carry a cost.

Business grade VOIP sessions can be rented. They are high quality, very reliable, and have almost zero lag. This is, essentually, commercial grade Vivox done over normal phones.

I use them every day in conferences all over the world, lag is not detectable. Everyone would use a good quality phone, or sip software with a high quality codec and dial in to a commercial grade conference server. One person would digitise the conference and send it to a shoutcast server for the audience. I'd envisage that technician would use SIP software on their computer and rebroadcast that. Again, linux using Jack - althought this time windows could do it too if they were very careful of other noises their computer makes.

Main problem would be quality. It would sound like a telephone call. Meh.

Cost would be in the order of $0.05 to $0.30 per line per minute.

Solution 3... Workable, lag free, low cost, likely ideal

Teamspeak. Millions of people in MMO dungeons can't be wrong.

A teamspeak server is very easy to set up. The software is stable, lag free and high quality. Each person would join the software with teamspeak, a technician would use linux and Jack, or less optimally windows to rebroadcast it.

https://www.teamspeak.com/en/

I'd certainly do some dry runs on teamspeak. I think it would work well.

If you need help with performing a trial let me know, I can do the technical stuff for you in advance.

Edited October 12, 2018 by Callum Meriman

[Love Zhaoying]

1 hour ago, Callum Meriman said:

There are two problems - One will be foldback, the performer can deal with this by removing their headphones as they speak, but that will be an issue with snappy dialog.

Is this like the term “feedback”?

[Callum Meriman]

4 hours ago, Love Zhaoying said:

Is this like the term “feedback”?

No, feedback is a screeching whistle. Foldback is the performer hearing themselves. It's a good thing in a live setting and quite desirable when it's instant, but when there is lag, it's really bad. Hearing your own voice 6 seconds delayed causes most people to stop talking, it's quite jarring.

Think of it like an echo on a cell phone, but worse.

These are live foldback speakers...

Edited October 13, 2018 by Callum Meriman

[Love Zhaoying]

7 hours ago, Callum Meriman said:

No, feedback is a screeching whistle. Foldback is the performer hearing themselves. It's a good thing in a live setting and quite desirable when it's instant, but when there is lag, it's really bad. Hearing your own voice 6 seconds delayed causes most people to stop talking, it's quite jarring.

Think of it like an echo on a cell phone, but worse.

These are live foldback speakers...

Here, we call those "monitor" speakers.

[Corwyn Allen]

On 10/12/2018 at 6:29 PM, Callum Meriman said:

Interesting problem! I can't speak from experience of doing it, although I stream a lot and can visualise some of the issues. I hope you don't mind me making these comments.

{SNIP}

If you need help with performing a trial let me know, I can do the technical stuff for you in advance.

Thank you Callum, that is precisely the kind of information I was looking for. Frankly I had given Teamspeak some thought and may have to try that out. One group I've been working with has been replacing Skype with Discord. Haven't really played with it a lot and don't know how much it differs from Skype with respect to lag between participants. Getting it in world is the big trick. And thank you for the offer of help. When I get to that stage in the process I may call on you. Thanks again.

Corwyn

[Corwyn Allen]

On 10/12/2018 at 8:07 PM, Love Zhaoying said:

Is this like the term “feedback”?

No, it's not feedback, which is what happens when your output feeds back into your input (speakers to mics) and it becomes a never ending loop of sound that builds to a brain shattering screech.

What Callum is referring to is what, when I was doing vidoeconferencing professionally, we called loopback. It's what you get when you go to test your Voice in SL at Voice Echo Canyon. Or what you get when you try to listen to your own stream being transmitted in world. A delay of anywhere between a couple seconds and 20 or so before you hear yourself. Can be maddening.

Corwyn