Add Voice recording functionality to train for Custom Neural Voice

[Yonet]

Depends on authentication.
Need to:

• Record voice in the required format.
• Create the text file for the recording detecting the sentence.
• Allow editing and deleting of the audio or text.
• Allow manualy typing the sentence for the utterence.
• Take a subscription and a resource to train the model
• Alternatively allow downloading of the formatted documents with audio and text for CNV training.

[Yonet]

@kartben FYI

[kartben]

Bonus points for providing user feedback with regards to the quality of their audio (i.e. flag early those recordings that would be flagged or rejected by the service anyway):

• Sampling rate is 24+ kHz
• Signal to noise ratio is 80 dB or higher
• etc.