Bug fixes with batch recognition

Joll59 / d-ser-t

d-ser-t quantifies speech recognition accuracy of the MSFT speech service and/or user created MSFT custom speech service models.

2 stars 2 forks source link

Closed zanawar closed 4 years ago

zanawar commented 4 years ago

Fixed 2 bugs with batch recognition:

The Internal Recognizer was skipping over every other file and crashing once it got to the end of it's file queue
The MultiFIlePullStream was sending stale data instead of silence