IssueTracker

Introduction:
→ C++ is a popular programming language for Speech-To-Text since it's the primary language used in Digital Signal Processing.

Problem Definition:
→ The Speech-To-Text StreamingRecognize* sample code in [1] uses a basic gRPC channel which opens up a single TCP connection to the Speech-To-Text backend.

Each HTTP/2 connection has a preset limit regarding the max ‘number of concurrent streams’ which is set at Google’s side. After the limit is passed, streaming RPC calls start to get blocked.

Feature Request:
→ The purpose of this FR is to track progress on the modification of the sample in [1] to use a custom channel  that pools streams across connections. This modification is needed to scale beyond the per connection stream limit.

[1] https://github.com/GoogleCloudPlatform/cpp-samples/blob/master/speech/api/streaming_transcribe.cc#L62
[2] https://github.com/googleapis/googleapis/blob/master/google/cloud/speech/v1/cloud_speech.proto#L71
*StreamingRecognize is a bidirectional RPC call [2] and is a good candidate for channel pooling