Wowza Community

DASH player architecture and tools

Hello

We are developing a new Player for my firm.

Today’s player uses win forms and uses a proprietary solution that implements all logic (streams synchronization, decoding etc) on the client side.

I am going to develop HTML5 based player and since I’m a new in this are I would like to consult with you which technology to use.

We record thousandth of telephone conversation between customers and call centers. Usual call duration is ~ 3-5 min (300-500K). We use telephone codecs 711 and 729 to keep these interactions (in a proprietary container).

We’d like our clients to have “Youtube” experience, so we cannot wait for the full file transcoding completed.

As a first step it should be deployed on the premise.

We think about MPEG DASH, transcoding on the server side from G711/729

to aac and MP4.

My question is if our direction is right and which tools can be used to transcode and convert files to standard formats?

Thanks

Efim

Hi Efim,

To clarify, your user requirements for your project are:

  1. Accept and record incoming streams (telephone conversations) from customers.

  2. Streams can be in the thousands.

  3. Stream duration is on average 3-5 minutes, 300-500K in size.

  4. Streams are currently encoded in 711 and 729 formats.

  5. Users should have live streaming experience, and cannot wait for full file transcoding to be completed (not sure what you mean by this; do you mean that the user can listen to a recording of their current conversation much like a DVR? or do you mean an archive sort of recording where they can look up past conversations?)

  6. Infrastructure needs to be self-hosted.

  7. Playback format will be MPEG-DASH.

  8. Streams need to be transcoded into AAC/MP4.

  9. The DASH player is custom.

Is this correct so far? Please do correct me if my assumptions are inaccurate. If the above is correct, then some of it would be possible with Streaming Engine. You can accept and record incoming streams and then serve these streams as VOD files similar to archive files. However, the biggest stumbling block is that Wowza Transcoder currently supports G.711, not G.729.

Michelle

Hi Efim,

To clarify, your user requirements for your project are:

  1. Accept and record incoming streams (telephone conversations) from customers.

  2. Streams can be in the thousands.

  3. Stream duration is on average 3-5 minutes, 300-500K in size.

  4. Streams are currently encoded in 711 and 729 formats.

  5. Users should have live streaming experience, and cannot wait for full file transcoding to be completed (not sure what you mean by this; do you mean that the user can listen to a recording of their current conversation much like a DVR? or do you mean an archive sort of recording where they can look up past conversations?) the conversation is stored and we want to [playback it in a few days, for example. We want to start playback during 2-4 sec, have an option to pause, seek. So we need to bring the call from the storage and transcode it

  6. Infrastructure needs to be self-hosted.

  7. Playback format will be MPEG-DASH. not must, we are looking for standard tech

  8. Streams need to be transcoded into AAC/MP4.

  9. The DASH player is custom.

Is this correct so far? Please do correct me if my assumptions are inaccurate. If the above is correct, then some of it would be possible with Streaming Engine. You can accept and record incoming streams and then serve these streams as VOD files similar to archive files. However, the biggest stumbling block is that Wowza Transcoder currently supports G.711, not G.729.

Michelle

Thanks Michelle,

See my answer

  1. Users should have live streaming experience, and cannot wait for full file transcoding to be completed (not sure what you mean by this; do you mean that the user can listen to a recording of their current conversation much like a DVR? or do you mean an archive sort of recording where they can look up past conversations?) the conversation is stored and we want to [playback it in a few days, for example. We want to start playback during 2-4 sec, have an option to pause, seek. So we need to bring the call from the storage and transcode it.

Since transcoding may take a while we need to start playback before it completed)

  1. Playback format will be MPEG-DASH. not must, we are looking for standard technology solution.

We are looking also to replace 729 with aac if it will be possible to reduce the file size.

So generaly speaking, the direction is right and wowza engine can be usable here?