I was watching some videos in twitch.tv to be sure how variants work for live streaming.
I see that, in the beginning a master manifest is downloaded and then a playlist is selected. After this, I used tomcast to throttling network and I waited until videos resolution changes to other variant, but it did not happened never (video was lagged but continued downloading the variant video selected in the beginning).
My question is, if live-HLS adapts the streaming with the variants once it started to play.
The selection of what variant to play is done by the player. Every player is different, and Twitch uses their own. To understand why that specific player is behaving that specific way would require Twitch to comment.
Related
I am just starting to play around with video.js and really like it. I currently have some code where I have two players showing two different HLS streams in a single browser page.
However, HLS inherently has high latency and that may not work for my project. So I am wondering if video.js can receive and play MPEG2-TS/UDP streams which would have less latency (I can easily change the format of all of my source video steams).
My basic requirement is to have 2 players in a single browser page, one player showing the video stream sent from a particular network node, and the second showing how a different network node received that same stream. So the two video.js players on the browser page are showing 2 video streams that are actually the same video so they are highly correlated. This is why the latency is a critical requirement for this project.
Thanks,
-Andres
What is required to use SMIL file to utilize adaptive streaming in a videojs player. I have created the SMIL file in my wowza application and it is creating my 4 separate streams and making them available. However I cannot get my webpage, that uses videojs, to correctly play the SMIL file. Hints on that coding or where to go to find the correct documentation would be greatly appreciated.
There aren't many implementations of SMIL players. I'm sure I've seen wowza URLs that suggest it will output the SMIL as other formats, something like whatever.smil/manifest.m3u8. That's HLS which could be played on mobile and Safari natively and with videojs-contrib-hls elsewhere.
I know the question is old, but I've been struggling with this recently, so I want to share my experience in case anyone is interested. My scenario is very similar: want to deliver adaptive bitrate streaming from Wowza to clients using videojs.
There is a master link that explains how to setup and run Wowza Transcoder for live streaming, and how to set up your Adaptive Bitrate Streams using an SMIL file. Following the video in there you can achieve to have a stream that uses ABS, but the SMIL file is attached to the stream name, so it is not a solution if you have streams that come to Wowza from another Media Server origin and that need to be transcoded before being served to the clients. In the article there are a few key things mentioned (like the Stream Name Groups), but somehow things doesn't seem pretty clear, at least to me. So here is some clarification from what I understood from all articles I read and what I did to achieve ABS:
You can achieve ABS in Wowza either with SMIL files or with Stream Name Groups (NGRP). NGRP refres to a block of streams that is defined in the Transcoder template that can be played back using multi-bitrate streaming (dynamically) (<- this is what I used). And SMIL files are used to create a "static" list of streams for multi-bitrate VOD streaming. If you are using Wowza Origin-Edge Delivery you'll need the .smil file, because NGRP do not get forwarded to the edge. (Source for all this information: here).
In case you need the SMIL file, you probably need to generate a new one for every stream, and probably you want to do that in an automated way, so best way would be through an HTTP request (in the link above it is explained how to achieve this).
In case you can live with NGRP, things are a bit easier:
You need to enable Wowza Transcoder (this is pretty easy and steps are in the video I mention above).
You should create your own Transcoder Template with the different stream presets you want to deliver, as an example you can check the default ones that are already there. The more presets you add, the more work Wowza will need to do whenever a stream comes, since it will need to generate a new stream for every preset that you have defined.
Now is when we generate the NGRPs. In your Transcoder Template, you can generate as many NGRPs as you want (to clarify: these are like groups of streams, that you will be able to set in your clients video player. Each NGRP contains the streams that the video will be able to use when doing the adaptive bitrate streaming). For instance, these are the default NGRPs:
If you play the ngrp "_mobile" in the clients video player, the ABS algorithm in the player will be able to adapt itself to play either the 240p or the 160p streams based on the client capabilities.
So imagine you have these two NGRP. In order to play them in videoJS, you will need to set the source to:
http://[wowza-ip-address]:1935/<name-of-your-application>/ngrp:myStream_all/playlist.m3u8
or
http://[wowza-ip-address]:1935/<name-of-your-application>/ngrp:myStream_mobile/playlist.m3u8
... based on how many options you want to provide to the client player to use for the ABS. (For instance: if your targets are old mobile devices, you probably just want to offer a couple of low bitrate streams).
(This would be in case you're delivering an HLS stream. If other format, the extension would change, for instance if you are delivering a DASH stream you would have "/manifest.mpd" instead of "playlist.m3u8").
That is all, there is also a very helpful link in video.js documentation explaining how it does the bitrate switching: here.
I hope it helps someone! At least clarifying things! :)
I want to create an application capable to play YouTube video's audios and also save the downloaded content in a local cache, therefore when the user decides to resume or play the video again, then it doesn't have to download part of video again but only download the remaining part (User can decide what to do with the cache then, and how to organize it).
It is also very convenient for mobiles (it is my main focus) but I'd like to create a desktop one too for experimental purposes.
So, my question itself is, does YouTube provide any API for this? I mean, in order to cache the download content I need that my application download the content and not any embed player (also remember that it is a native application). I have a third-party application in my Android system that plays YouTube videos, so I think it's possible unless that the developers use some sort of hack, again this is what I don't know.
Don't confuse with the web gdata info API and the embed API, this is not what I want, what I want is to handle the video transfer.
As far as I know, there is no official API for that. However, you could use libquvi to look up the URLs of the real video data, or you could have a look at how they do it and reimplement it yourself (see here).
I was doing some research to find out ways that would allow me to stream music on my website legally. I came across iTunes partner program which allows to stream music on a website through their embedded players. I was wondering is it possible to stream iTunes music through your own custom player? If that is not possible via iTunes, then what other methods are available?
You could do this with a server software like Icecast, there is some good tutorials on setting this up here: http://www.icecast.org/docs.php
Depending on how many browsers you want to support you might want to setup two streams, one in MP3/OGG and a "backup" stream in Flash. Then add some detection as to what the browser supports and present the correct stream (i.e.: Use the HTML5 <audio> tag for playing MP3/OGG to browsers that support this, and use your flash stream for the rest)
their program allowing playback of music in the iTunes Store is likely only for those with the intention to sell music, without providing a commerce business, you'd be breaking their partner program T&C's.
Lots of ancient, non-negotiable history due to mergers and acquisitions, so I realize there are better ways to do all of this, however... I am faced with the following:
I have an osmf based video player where a particular playlist item (for a live video), must do the following:
play a preRoll prior to displaying a countdown
display a countdown until the video start time (synched with the server time)
play another preRoll prior to video playback
play the live stream until the server time reaches the stream end time
then play a post roll video
I've gotten this working for the most part, but I'm running into walls with the arbitrary insertion of "ads", since I don't want to trigger events associated with loading new media. If I try to inject a new ad (particularly after the stream has played), the live stream will display again. While I could figure out some horrible way to make this work, I just wanted to make sure first that I'm not missing something critical about osmf and live events. I'm also a bit uncertain as to what is native osmf in the architecture I'm working with and what is homegrown.
1) Does osmf have a concept for a scheduled live event that might make this easier
2) Does osmf have an option to arbitrarily insert a video into playback based on some external call without changing the playlist index or returning to the beginning of the video.