A Buyer's Guide to Format Conversion Tools
The confusion often lies in the fact that all but one type of ABR format -- the Apple HTTP Live Streaming (HLS) format -- use fMP4, but each uses a different manifest file. HLS, which uses MPEG-2 Transport Stream (M2TS), has a manifest file with an .m3u8 file name extension, while others use extensions such as .ism or .f4m. The emerging MPEG DASH standard uses another extension altogether.
Transmuxing solutions for ABR (e.g., HDS to HLS) convert one of several fragmented-MP4 streams, each containing either H.264 (AVC) video or AAC audio, into a single transport stream (.ts) file for segmentation into the proper manifest file so that the proper chunk or segment is available -- at the optimal bitrate -- when an Apple device requests it.
Changing Your Aspect (Ratio)
Another question to ask is which format is best to start with, along with the corollary regarding what the transcode end result will look like.
For instance, in the analog television era, a live PAL television signal might have been converted to NTSC or SECAM for broadcast across borders or geographies. For live events, this format conversion meant converting both the frequency and resolution of the signal. In the world of streaming, though, not only would the frequency (codec) and resolution (pixel size) need to be changed, but the bitrate and even the aspect ratio often need to be changed as well.
Aspect ratio is one area that's remained constant as we've moved from analog to over-the-air (OTA) transmission. During the widespread adoption of HDTVs, format conversion also required aspect ratio changes, from 4:3 legacy televisions to 16:9 modern televisions. These digital format convertors are still a large part of OTA delivery, and they can be used in the world of live online streaming video when it comes to moving content to mobile screens that may not have an aspect ratio that's either 4:3 or 16:9.
Number of Inputs
Format conversion, in order to deal with the computationally heavy preprocessing, will convert one signal type to another signal type. On the other hand, those live transcoding solutions that have IP inputs often have enough horsepower to handle multiple transcodes of multiple signals.
In an ideal ABR setup for today's broadcast-to-TV-everywhere environment, a single stream will enter the live transcoder, and the output will be several outputs at varying -- but consistent, in terms of aspect ratio -- discrete bitrates.
If all content is delivered to the transcoder in H.264 as a single IP stream -- typically M2TS via a .ts packet -- then the process of converting is known as transrating because only the bitrate will change. Even if the resolution also changes, it's also referred to as transrating as long as the aspect remains constant.
Content delivered, in our previous scenario, using the MPEG-2 codec, however, would require transcoding, as the codec itself needs to be converted from the MPEG-2 codec to the H.264 codec. The same is true for content that is encoded in H.263, the predecessor of H.264, or Flash 8 Video or WebM.
One example of this would be a live transcoding box that receives in M2TS, demuxes the content, transcodes the content into H.264, and then repacketizes it into another MPEG-2 transport stream for retransmission.
Live transcoding solutions abound, especially for converting from MPEG-2 codec content, wrapped in a transport stream, to H.264 content, suitable for segmentation for various flavors of adaptive bitrate manifest files. Determine first whether you need live transcoding, then determine whether transrating, transmuxing, or transcoding are needed. Finally, compare offerings that not only convert from one codec to another but also provide segmentation and packaging for today's newer HTTP-based delivery options.
This article appears in the forthcoming 2013 Streaming Media Industry Sourcebook.