Save up to 20% Bitrate for Live Streaming with Content Adaptive Encoding
by Prashanth Dixit (Principal Engineer, Media Server Technologies)
Live streaming has become ubiquitous, with all major service providers of TV Everywhere, OTT streaming and social media networks offering live audio video feeds. When it comes to video quality, consumers today expect nothing but the best. The margin for error is even lower with live streaming. While the primary challenge is to ensure live video is not choppy and don’t drop frames, service providers are also confronted with this overwhelming challenge: “How do I bring down my operational expenses and effort in delivering an outstanding QoE?”
Advanced Content Adaptive Encoding (CAE) technology provides the answer.
As explained in our previous blog, leading video streaming providers have realized the benefits of CAE to deliver outstanding OTT/VOD content. However, traditional CAE solutions might not be well positioned to tackle some of the unique challenges presented by live streaming in terms of processing overhead and latency. Therefore, what you need to ace live streaming is THINKode Live – Ittiam’s ML based, non-iterative CAE solution that helps fulfill the key requirements of latency-critical, HTTP based live streaming applications.
‘Unlike traditional multi-pass encoding with fixed bitrate ladder, CAE solutions allocate only the required bits for a given video, based on its complexity, to drive significant bitrate savings.’
Tackling Live Streaming Challenges with Content Adaptive Encoding
Let us take a look at some of the key challenges with traditional CAE solutions and how service providers can benefit from moving to a more advanced CAE solution such as Ittiam’s THINKode Live for live streaming.
1. Processing Cost to Achieve Bitrate Savings
Traditional CAE solutions are iterative in nature and thus processing intensive. Online quality measurements, often a part of these iterations, is a major part of the processing costs. However, if you choose a solution that implements a non-iterative one-shot prediction scheme, you can overcome this hiccup with ease.
Ittiam’s THINKode Live uses such an ML model trained offline with thousands of clips. It predicts both the optimal resolution and bitrate for a given input video with as high as 90% prediction accuracy. All this while eliminating the need for online quality measurement and utilizing only marginal additionalprocessing costs.
Driven by a non-iterative one-shot prediction model, THINKode Live enables
up to 20% savings in bitrate, and hence huge savings in OpEx
2. Additional Latency for Pre-Analysis
Latency is the other crucial factor in live streaming. Traditional CAE solutions pre-analyze content by looking ahead into the video (either the entire content or a segment). This adds to the encoding workflow latency, which can make it an unviable option for service providers.
The challenge for the CAE solutions is to minimize this latency overhead for live streaming. For instance, THINKode Live can start predicting optimal encoding parameters within two seconds, while continuing to tune the prediction accuracy with further analysis.
THINKode Live adds <2 seconds latency to the encoding workflow – therefore a great choice for
service providers looking to stay ahead in the live streaming ecosystem
3. Codec Dependency for Deployment
Some of the existing CAE algorithms reside within a specific codec implementation, making upgrades to existing workflows necessary to leverage their benefits. This is fraught with high time and effort costs, needing significant expenditure on revalidation and deployment of the new encoder.
THINKode Live however is codec independent as it sits outside the codec, providing you great flexibility to update existing workflows with ease.
THINKode Live offers you the flexibility to extend the life of existing H.264 encoders by delivering almost 80-90% of the bitrate savings promised by next-gen encoders
Redefine Live Streaming with THINKode Live
Ittiam’s CAE solutions have been proven to deliver impressive gains with VOD workflows. The challenges in achieving similar benefits for live streaming are different, something we have addressed in THINKode Live – an optimized CAE solution particularly suited for latency-critical live streaming applications. THINKode Live enables up to 20% bitrate savings at very affordable processing costs, while adding <2 seconds latency to the encoding workflow. By leveraging this codec independent solution, service providers can also enjoy the certainty of continuing with their existing codec implementation.