import podonosfrom podonos import *client = podonos.init(api_key="<YOUR_API_KEY>")etor = client.create_evaluator( name='Naturalness of my new speech synthesis', type='NMOS', lan='en-us')# Add a file.etor.add_file(file=File(path="<YOUR_AUDIO_FILE_PATH>", model_tag="my_model", tags=["syn1, epoch2"]))etor.close()
That’s it! This code will enable you to evaluate the naturalness of US English in your audio files in 5 point MOS (Mean Opinion Score).Once you finished running the code, you will see a draft evaluation in your Workspace.
In each evaluation page, you can see the type, scale, language, and many more. Once you confirm everything is correct, you can kick off the evaluation by clicking Pay and request evaluation button at the bottom. Also, do not forget registering your payment method.
Once the payment is complete, your evaluation request is confirmed. You can view the details of your evaluations from the Workspace or through email.
After the evaluation finishes, you will have an access to the evaluation report.Of course, you can configure more details of the evaluation setup. In Naturalness, we will go through the details under the hood.