LiveKit Community

hey steve, the <https://github.com/livekit/client-sdk-js/blob/main/src/room/track/RemoteTrackPublication.ts#L130|logic is here> for automatically managed layer switching. The server has logic to automatically match the requested resolution to the right layer.

currently it's a fixed two lower layers at 180p and 360p. we don't support customizing it at the moment.

Thanks <@U01KZ61GV5Z> Can a Client re-subscribe to a particular "layer" of publisher if it wanted to ? Says client knows that this publisher is on thumbnail and no point getting the highest resolution but 180p or 360p one, can client switch/re-subscibe to that low-dim layer and how to do it? Thanks.

you can do it manually via the api described here in the docs: <https://docs.livekit.io/guides/room/receive#simulcast-controls>

Ah, very smart :slightly_smiling_face: Thanks <@U02EU2ZPTQU>

Hello :wave:, can you release a new version for livekit react we need to use the changes

What language are you using? Can you provide some sample code where you notice this behavior?

Oh it is a long codes base inside a web-app. We simply use simulcast with setting likes this:
'on_stage': {
        'width' :  { 'min': 640, 'ideal': 1280},
        'height':  { 'min': 360, 'ideal': 720 },
        'frameRate': {
             'min': 24,
             'ideal': 24,
             'max': 30,
        },
        'facingMode': { 'ideal': "user" },
        'max_bitrate': 1500,
        'codec': LK_VIDEO_CODEC_DEFAULT,
        "quality": "HIGH",
        "simulcast": True,
    }.

and for AUDIO:
    'on_stage' : {
        "autoGainControl": True,
        "channelCount": 2,
        "echoCancellation": True,
        "noiseSuppression": {
            "ideal": True,
        },
        "sampleRate": 48000,
        "sampleSize": 16,
        "max_bitrate": 128,
        "latency": 0,
        "codec": LK_AUDIO_CODEC_DEFAULT,
        "simulcast": True
    },

On iPhone and Android it is heatingly hot and quality is soso and bad even on Wifi, 4G it can not even start streaming. The same setup works just fine on all PC's browsers.

we haven't seen this on mobile using the sample app. I used 720p and simulcast on iPhone 12 (using vp8 to trigger software encoding), it appears to be working fairly well. The phone does get a bit warm, but not out of the ordinary.

as to the performance issues on 4G. this is due to Safari's handling of limited bandwidth. Chrome handles this extremely well, and we haven't spent a ton of time to see how we can work around safari limitations.  In the mean time, I recommend sending lower resolution/bitrate streams coming from Safari.

Reduce max_bitrate to 600 works OK OKin 720p w/ simulcast, a bit shaky but that is fine. Thanks,

This is one of the BEST open source team I have ever met. Open-minded, extremely knowledgable, passionate and available to support community. Thanks, please keep it up.

thank you guys! really appreciate the kind words :pray:

Hi all, great to find this community. I run FRAME (<https://learn.framevr.io>) which lets people come together inside a 3d space as avatars to communicate and collaborate. We're thinking of using Livekit and would love to hire someone to help us explore and possibly map out the integration. Any takers? :slightly_smiling_face:

hey alex, 50 participants shouldn't be an issue.. though you'd want to use <https://docs.livekit.io/client-sdk-js/interfaces/connectoptions.html#automanagevideo|autoManageVideo> to enable an optimization that pauses video elements that aren't visible.

FWIW, in our most recent bench against <http://example.livekit.io|example.livekit.io>, we had 100 participants running in a room, with each subscribing to all other participants. This was done on a 16 core machine. At peak it was at 65% CPU utilization

Somewhat tangential question for people here - has anybody on Flutter had experience with playing back RTMP? We need to be able to play it back to supplement LiveKit rooms in cases where end users are originating streams on other platforms

Hello
How many participants can I have (in a single meeting) with a VM (1 vCPU x 1thread, 1 GB RAM) ?

What if I want to have a room with 35-40 participants, but only 6 of them are visible?

hey <@U02RT1M3P6U>, you can benchmark yourself (ideally with your app) to determine participant scalability, as it's highly dependent on the use case. our benchmarks are published <https://docs.livekit.io/deploy/benchmark#video-room|here>.

with `autoManageVideo`, you only need to worry about layout, ensuring only speaking participants are visible. it would optimize bandwidth use automatically.

Not visible element means "visibility:hidden" or  "display:none" CSS rule?

any of the above.. not visible to a user in the browser view port

6 video streams + 6-8 audio = max 14 streams sent to each participant
So, that's 14*35 = 490 streams per 1 room

:wave: Hi everyone!

QUICK QUESTION: Does livekit have active speaker detection. as a way to implement active speaker animation on the UI side

I think this comes up often and there has been interest by both end-users of livekit and the core team but I believe react-native client wasn't able to be implemented because react-native-webrtc hasn't implemented some necessary features.

Most recent update I've seen is this issue: <https://github.com/react-native-webrtc/react-native-webrtc/issues/851>

LiveKit team had previously advocated for this PR which wasn't merged: <https://github.com/react-native-webrtc/react-native-webrtc/pull/773>

I guess the alternative would be to build on top of the ios/android libraries instead of rn-webrtc?

If you go that route, you could try writing most of your app in react-native and then use native views to call the ios/android libraries  <https://reactnative.dev/docs/0.61/intro-react-native-components>

but take that with a grain of salt, as I've not done much with react native myself.

Thanks, I’ve written a few RN libs so I could probably try this, just wanted to check if it’s already being worked on

hi <@U02R6M4K326> it'd be great to have a react-native lib! We can definitely try to wrap our native libs. If you are interested in contributing to this, lmk!

I might have some time to look at it during the holidays but can’t promise anything haha (startup priorities…)

I am having trouble getting the server started and tested

Can you post what errors are you getting? It’s difficult to help you without much context.

And if you believe it’s a bug in the server, please file an issue <https://github.com/livekit/livekit-server/issues>

config.yaml

Hi Peter,  I installed the server via the direction in the documentation. This install was completed on a GCP VM , I opened all ports on the firewall so I can test with no issues. The server is built using Ubuntu ""18.04.6 LTS "  and running livekit with the following yaml

`LIVEKIT_KEYS="APItcRyceNwouTb: g346fAENVRlj70UPRSHmeg8v1EhvZk3l" ./bin/livekit-server --config=config.yaml`

`root@livekit-try2:~/livekit-server# LIVEKIT_KEYS="APItcRyceNwouTb: g346fAENVRlj70UPRSHmeg8v1EhvZk3l" ./bin/livekit-server --config=config.yaml`
{"level":"info","ts":1639956182.4707496,"logger":"livekit","caller":"logger/logger.go:26","msg":"using single-node routing"}
{"level":"info","ts":1639956182.4763606,"logger":"livekit","caller":"logger/logger.go:26","msg":"Starting TURN server","turn.portTLS":443,"turn.externalTLS":false,"turn.portUDP":443}
{"level":"info","ts":1639956182.4776542,"logger":"livekit","caller":"logger/logger.go:26","msg":"starting LiveKit server","addr":":7880","nodeID":"ND_h8D8qydm","nodeIP":"34.139.68.171","version":"0.14.2","rtc.portTCP":7881,"rtc.portUDP":7882}

when I run tests against the server i always get connection failed

I don't see any  errors in the logs , so I am out of ideas

I was able to get "<https://github.com/AirenSoft/OvenMediaEngine>" to work but not livekit , I am trying to see what will be best for my project

root@livekit-try2:~/livekit-server# ./bin/livekit-server --key-file keys.yaml create-join-token --room "myroom" --identity "myidentity"
Token: eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJleHAiOjE2NDI1NDg4MTIsImlzcyI6IkFQSXRjUnljZU53b3VUYiIsImp0aSI6Im15aWRlbnRpdHkiLCJuYmYiOjE2Mzk5NTY4MTIsInN1YiI6Im15aWRlbnRpdHkiLCJ2aWRlbyI6eyJyb29tIjoibXlyb29tIiwicm9vbUpvaW4iOnRydWV9fQ.oDHpyUmVeY2kX8lLs6gVFq23lcTQy-svTrJtS09k4VM

If I do a tcpdump "tcpdump -nnSX port 443" i see traffic hit the server using your <https://example.livekit.io/>

Do you have an SSL cert setup? Please see <https://docs.livekit.io/deploy/vm|docs here>. Browsers disallow connections to insecure HTTP servers (from an already secure endpoint)

yes I have ssl cert setup and configured in my yaml

that's for the TURN server. what about for the HTTP port? (7880)

i can create another cert , where do I put that pem file

I assumed the turn server takes in 443 using cert and routes to 7880 having already done the ssl negotiation

livekit doesn't terminate SSL for http, since any load balancer, nginx or haproxy could do it

ok so I have to make a nginx ssl offloader to port 7880?

yeah, our docs talks about SSL &amp; domain requirements too: <https://docs.livekit.io/deploy/#domain-ssl-certificates-and-load-balancer>

Today I got a similar error like yours "server unreachable" after following the official tutorial and after few hours of investigation I found the issue with the YAML init template INDENTATATION (When I paste the SSL certificate it was not indented )

I jumped the gun on Ruby — one exists but it’s not public yet. :smiling_face_with_smiling_eyes_and_hand_covering_mouth:

<@U0291E7JEP3> led charge on the Ruby SDK and I think it's looking quite good. We should prep it for release sometimes soon :slightly_smiling_face:

cool i just needed access token generator for now which was pretty trivial to write but sounds like you have the rest already underway

Welcome <@U02S132K2UQ>! Thanks for giving us a shot :slightly_smiling_face: Let us know if you run into any issues or have any other feedback.

What were you using before? asking out of curiosity.

Thanks. Been using Jitsi, but not a ton of experience with scaling. Was reading experiences of both and overall this looks like it's on the right path.

external TLS termination &lt;&lt;&lt; is that for Turn??

With v0.15.1 you can also upgrade the recorder to v0.3.12, which includes compatibility options for h264 encoding profiles
<@U02K46F1YLW> <@U02HN8GPUBY> <@U02Q369F4BF>

oh, I gotta upgrade asap! :slightly_smiling_face:

<@U02HN8GPUBY> yes it is! let me clarify in release notes.

"Feel likes" it handles congestion better... But lot more tests are required though.

try turning off dtx when publishing the track.

Where do I do that <@U01KZ61GV5Z>?!?
Also where can I force to replace the below  SDP in LiveKit ReactKS code ?
```sdp.replace('useinbandfec=1', 'useinbandfec=1; stereo=1; maxaveragebitrate=510000');```
Thanks much.

no, it's one of the publish options: <https://docs.livekit.io/client-sdk-js/interfaces/trackpublishoptions.html>

livekit doesn't support munging of SDP

which option are you trying to set? perhaps we can expose that via the SDK as an option

Please do so, to make AUDIO quality good enough we would have to munge SDP sometime <@U01KZ61GV5Z> Now I know that whatever setup we do, its audio will be PHONE-quality only :disappointed: We will need to replace the standard MONO into STEREO and a few more.

you can convert a video or mp3 file with ffmpeg
```ffmpeg -i someMusic.mp3 -c:a libopus -page_duration 20000 -vn someMusic.ogg```
then you can try with the livekit-cli to publish the .ogg file
```livekit-cli join-room --url <ws://localhost:7880> --room yourRoomName --identity someId --api-key yourKey --api-secret yourSecret --publish someMusic.ogg```
I just tried it and it sounds great.

Is stereo the option you are trying to set? are there anything else?

<@U02HN8GPUBY> is right, if you can programmatically publish the track to livekit directly, that'd be better than trying to mess with browser encoding options.

Nahhh <@U02HN8GPUBY> this is Music Theatre application, all LIVE mate :smile: We have to LIVE mixing voice + background music (something like Karaoke), that is the intention.

<@U01KZ61GV5Z> I believe that the quality of the audio is extremely important and being able to publish truly STEREO music is something we should be able to do. Something like the below will help much.

sdp = sdp.replace('useinbandfec=1', 'useinbandfec=1; stereo=1; maxaveragebitrate=&lt;max_bitrate*1000&gt;');

Thanks.

<@U01KZ61GV5Z> if possible, we should be able to try  trial/error the option below also to completely avoid variable bitrate for audio, and instead of
```a=fmtp:111 minptime=10; cbr=1```
...meaning either 2 options:

sdp = sdp.replace('useinbandfec=1', ''useinbandfec=1; stereo=1; sprop-stereo=1; maxaveragebitrate=&lt;max_bitrate*1000&gt;');
OR
sdp = sdp.replace('useinbandfec=1', 'cbr=1; stereo=1; sprop-stereo=1; maxaveragebitrate=&lt;max_bitrate*1000&gt;');
in SDP as well. If possible let us try these both.

We are "munging" the LiveKit SKD client, and will test. Will share the code changes here.

The nodejs server sdk documentation is at <https://docs.livekit.io/server-sdk-js/>

The main thing you’ll need to get going is to generate access token (also explained in the main docs <https://docs.livekit.io/guides/access-tokens>) . Then you’ll need to send the token to the client (how you do that is up to you but genearlly it goes into your rest or graphql api)

As a final step, you’ll use the token on the client by passing it to the Room <https://docs.livekit.io/#client-examples>

<@U02R6M4K326> i have my backend  running on localhos:7000 and i tried to create a room as shown below
```const livekitHost = '<http://localhost:7000>';
 const svc = new RoomServiceClient(livekitHost, 'api-key', 'secret-key');
// create a new room
const opts = {
  name: 'myroom',
  // timeout in seconds
  emptyTimeout: 10 * 60,
  maxParticipants: 20
}
svc.createRoom(opts).then((room) =&gt; {
  console.log('room created', room)
})```
but  i got this error  when i run the server.                                                                                                     (node:41383) UnhandledPromiseRejectionWarning: Error: Request failed with status code 404
    at createError (/Users/Username/Google Drive/CityHi/Server/node_modules/axios/lib/core/createError.js:16:15).                                                                                                         am not really sure why, or dose it has to do with the  localhost ?

the livekit host should be `<ws://localhost:7000>`, not http

<@U02R6M4K326> Am still having this same error,   the setup i have can bee seen on this gist  <https://gist.github.com/osain-az/b1b009f30455772366d92aa141002311>

You actually need to run LiveKit, see <https://docs.livekit.io/guides/getting-started#start-the-server>

the nodejs server sdk enables your backend to communicate with the LiveKit server but the LiveKit server itself needs to run in a separate docker process

<@U02R6M4K326>  Thank you for your quick response.  the setup is done and i can establish  connection from the server/backend.     Merry Christmas

Hi Steve, are you specifically seeing this happen in Mobile Safari? There’s a bug right now in iOS15 causing audio to be routed to the earpiece instead of the device speaker.

<https://bugs.webkit.org/show_bug.cgi?id=230902>

I think someone who tried this with a web version I had built from a very slightly modified version of the client-js example had the same issue on android. Bluetooth wasn't an option for the audio output. I think he then tried building the flutter app and had issues with it there too but managed to make some changes to get it working. Sorry, I don't have more details since it was just some quick testing that we haven't followed up on yet.

<@U01KZ0SCDKL> on IOS 15 Safari &amp; Chrome. But YouTube on browser is louad and also push to bluetooth speaker OK...

<@U01KZ0SCDKL> YouTube play on browser of mobile or tablet output audio nicely to speaker or buetooth nicely. That is a surprise. YouTube can but we can not

AFAIK, YouTube isn't using WebRTC for their audio playback.

here is a little trick to get around the bug, instead of putting the whole stream into an video element, grab both video and audio tracks, then put the video into the video element, then create a audioContext for the audio, then your speaker will work

you can try to build for windows using the latest client-sdk-flutter, it seems to work

as an alternative, you can build a web release using flutter (or javascript) and then build an electron or nw.js app, I've tried both methods, nw.js seems to use less cpu%

Hi i run the latest client-sdk-flutter, and build the example ,it will show lost connection to device

can you show us how you are setting the api keys? basically I just run
```docker run --rm livekit/livekit-server generate-keys```
to generate a key and set it in the config.yaml like this
```keys:
  key1: "APIVGVAsreDyEn2: pqErUWj92VSnvEKOupGUSngBvhGgBn9B1nCvQpA6V9E"```


hey <@U02S1919YLB>, I appreciate the note. we do have  docs on the config format: <https://github.com/livekit/livekit-server/blob/master/config-sample.yaml#L56>

we should probably link to it from the getting started guide.

Actually, ignore what I said.  It's my bad.  At the time I didn't quite grok what was needed.  Now of course going through things and it clicking.  Now I get it.  So yeah, got it working through the config.

Screenshot 2021-12-27 at 19.36.01.png

this looks awesome!

we use redis as it's used both as a messaging bus (pub/sub), as well as a data store. there are no plans to use another DB at the moment, but it's abstracted so that you may choose to swap it out.

Thanks!

Also thanks for the reply.

I was hoping it was just a data store.  But when the time comes I'll take a look and if I can, I'll make a PR.

Cheers!

<@U02S1919YLB> Your discord clone looks really  great! I am interested to start using livekit framework for voice and chant as well and i am looking for suggestion about how to start.
I see use cases for video conferencing but nothing related to chat.

If you're not using the react lib you can create your own video element and pass that ref to the attach