Using GPU to accelerate Speech Encoders/Decoders


I am considering to use GPU for accelerating Speech Encoder/Decoders like Opus , G.729, AMR WB etc…
Currently these Encoders/Decoders run on a CPU.
Is there any written documentation of performance that can be achieved, experience with this issue, libraries etc…

Thanks All

Not sure if you can find useful info in this thread.