Example of Infinite Streaming for Google Speech-to-Text API v2

[rodrigoGA]

I have not found any examples of Infinite Streaming with the Google Speech-to-Text API v2, and it's unclear whether there is a mechanism in version 2 of the API that facilitates this functionality. Could you provide an example of implementation or confirm if this capability is supported?

Thanks!

[holtskinner]

I found this code sample for endless streaming from microphone input using V1.

Linked in this documentation: https://cloud.google.com/speech-to-text/docs/endless-streaming-tutorial

https://github.com/GoogleCloudPlatform/python-docs-samples/blob/main/speech/microphone/transcribe_streaming_infinite.py

This code sample should work mostly the same for V2, you would just need to swap the imports from v1 to v2. I believe the client methods are the same.

[rodrigoGA]

No, the code is not interchangeable. The way to start the client, the API response, and I suspect the way it operates have changed.
I think a functional example would be useful since it's one of the most common use cases for the API.
Here is a code based on the v1 example, but it doesn't work as it should: https://gist.github.com/rodrigoGA/644ab63d244a9f5674e935540c6db6da

Here is a basic example I am working on, not based on the infinite streaming of v1, which does work, but I haven’t finished it yet. There are things I still don't understand about the library; I find it quite unfriendly to use. Version 1 was already not user-friendly, but version 2 is even more so.
I plan to create a wrapper around MicrophoneStream so that the Resumable streaming code can be reused outside of this example. I greatly appreciate any guidance https://gist.github.com/rodrigoGA/020a144cbd239ed0d9eba6e44c8682c0

[gangchen03]

Submitted a PR #11847
it should fix the example request.

[rodrigoGA]

Thank you @gangchen03 it was helpful.
But the example doesn't seem to work continuously.
After having the microphone active for 5 minutes with the telephony model, the following error message is received:

Traceback (most recent call last):
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/google/api_core/grpc_helpers.py", line 116, in __next__
    return next(self._wrapped)
           ^^^^^^^^^^^^^^^^^^^
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/grpc/_channel.py", line 543, in __next__
    return self._next()
           ^^^^^^^^^^^^
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/grpc/_channel.py", line 969, in _next
    raise self
grpc._channel._MultiThreadedRendezvous: <_MultiThreadedRendezvous of RPC that terminated with:
	status = StatusCode.ABORTED
	details = "Max duration of 5 minutes reached for stream."
	debug_error_string = "UNKNOWN:Error received from peer ipv4:142.251.134.42:443 {grpc_message:"Max duration of 5 minutes reached for stream.", grpc_status:10, created_time:"2024-06-10T16:55:08.43835454-03:00"}"
>

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/home/rodrigo/Proyectos/marta/Call-Services/test/performance_test/testReasumableGoogleTranscribeV2.py", line 389, in <module>
    main(project_id)
  File "/home/rodrigo/Proyectos/marta/Call-Services/test/performance_test/testReasumableGoogleTranscribeV2.py", line 368, in main
    listen_print_loop(responses_iterator, stream)
  File "/home/rodrigo/Proyectos/marta/Call-Services/test/performance_test/testReasumableGoogleTranscribeV2.py", line 241, in listen_print_loop
    for response in responses:
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/google/api_core/grpc_helpers.py", line 119, in __next__
    raise exceptions.from_grpc_error(exc) from exc
google.api_core.exceptions.Aborted: 409 Max duration of 5 minutes reached for stream. [type_url: "type.googleapis.com/util.StatusProto"
value: "\010\n\022\007generic\032-Max duration of 5 minutes reached for stream.*8\013\020\206\326\215\'\032/\022-Max duration of 5 minutes reached for stream.\014"
]

and the script stops working.

It also doesn't work correctly with smaller models like telephony_short, which I understand are the desired ones to use in this scenario.
In these cases, when a sentence is transcribed as final, the following error result occurs:

0: NEW REQUEST
9240: Hola

240000: NEW REQUEST
Traceback (most recent call last):
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/google/api_core/grpc_helpers.py", line 170, in error_remapped_callable
    return _StreamingResponseIterator(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/google/api_core/grpc_helpers.py", line 92, in __init__
    self._stored_first_result = next(self._wrapped)
                                ^^^^^^^^^^^^^^^^^^^
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/grpc/_channel.py", line 543, in __next__
    return self._next()
           ^^^^^^^^^^^^
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/grpc/_channel.py", line 969, in _next
    raise self
grpc._channel._MultiThreadedRendezvous: <_MultiThreadedRendezvous of RPC that terminated with:
	status = StatusCode.INVALID_ARGUMENT
	details = "Audio chunk can be of a maximum of 25600 bytes. Received audio of 294400 bytes instead."
	debug_error_string = "UNKNOWN:Error received from peer ipv4:142.251.133.10:443 {created_time:"2024-06-10T17:01:02.223753051-03:00", grpc_status:3, grpc_message:"Audio chunk can be of a maximum of 25600 bytes. Received audio of 294400 bytes instead."}"
>

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/home/rodrigo/Proyectos/marta/Call-Services/test/performance_test/testReasumableGoogleTranscribeV2.py", line 389, in <module>
    main(project_id)
  File "/home/rodrigo/Proyectos/marta/Call-Services/test/performance_test/testReasumableGoogleTranscribeV2.py", line 365, in main
    responses_iterator = client.streaming_recognize(
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/google/cloud/speech_v2/services/speech/client.py", line 1641, in streaming_recognize
    response = rpc(
               ^^^^
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/google/api_core/gapic_v1/method.py", line 131, in __call__
    return wrapped_func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/rodrigo/Proyectos/marta/Call-Services/venv/lib/python3.11/site-packages/google/api_core/grpc_helpers.py", line 174, in error_remapped_callable
    raise exceptions.from_grpc_error(exc) from exc
google.api_core.exceptions.InvalidArgument: 400 Audio chunk can be of a maximum of 25600 bytes. Received audio of 294400 bytes instead.