CLI tool that measures latency of cloud streaming speech-to-text APIs. It simulates a realtime audio stream and primitive speech command parsing to measure the amount of time required for an API to provide the client with enough responses to begin actual command processing like playing a song or ordering a burrito.
There are plans to eventually support latency measurements for all major STT vendors including Google Cloud, AWS, Azure.
Set location of your project credentials file to GOOGLE_APPLICATION_CREDENTIALS enrivonment variable. Then run python cli.py. Available options are:
-i, --input TEXT Audio WAV file
-p, --pattern TEXT Pattern to match the transcript against, string or a file path
-l, --language_code TEXT Language code, defaults to 'en-US'Example command produced using speech synthesis is provided in the sound folder. Command to run the program would then look like this:
python cli.py -i sound/book-tickets-en.wav -p "book tickets from Seattle to Tokyo on June 7th"