Why Do I need a Persistent Optimization Process?

Any speech application will degrade in recognition performance over time due to a variety of issues such as new content or changes in the way callers refer to a subject. Persistent Optimization provides a regularly scheduled method for updating a voice recognition application without impact to callers.

What is SpeechCycle “persistently optimizing?”

Speech recognition accuracy supported by content, (the words and phrases that will be recognized), voice user interface design, and speech models which make up the foundation of converting sound to meaning are covered. SpeechCycle captures thousands of caller utterances from each application with great frequency. These utterances are transcribed and analyzed to ensure the speech application and caller expectations are in alignment. Any statistically significant performance issues are addressed by modifying the grammar, changing prompts, or adjusting speech recognition parameters within the scope of the application.

What is the difference between Persistent Optimization and Speech Tuning?

Frequency, cost, and coverage of all key elements of each speech application are the primary differentiators. Every organization deploying a speech application will tune it, especially in the deployment phase. Tuning however then becomes a discreet event, usually sporadically scheduled with a high cost associated to cover the expensive participation of voice user interface designers and speech scientists (if they are even available). SpeechCycle on the other hand programmatically optimizes each application. Processes are in place for utterance capture, transcription and analysis and modification.

How does SpeechCycle’s accomplish Persistent Optimization?

Established processes and unique tools make the Persistent Optimization Process a well established procedure. Transcription and analysis tools speed the process of pinpointing areas for improvement as well as quickly making changes.

What effort is required by me to support the Persistent Optimization Process?

SpeechCycle has the staff, tools, and process in place to take care of everything “behind the scenes”. But, it is usually not a good idea to surprise callers with new equipment or marketing programs without initially providing support in the speech application for the new words, phrases, or call flows that might be associated with change. Effort generally amounts to establishing clear communication to provide SpeechCycle with detail on any new equipment, programming, or marketing programs as early as possible.

How do I know that it is working?

SpeechCycle’s web-based Business Intelligence Portal provides granular statistics to quickly assess performance.