AI-generated summary
Genesys Cloud's dictionary management feature enables users to enhance transcription accuracy by adding organization-specific terms into the system to ensure proper recognition of names, acronyms, and industry language in transcripts. Users can access dictionary management through Admin > Quality > Dictionary Management or Menu > Conversation Intelligence > Speech and Text Analytics > Dictionary Management to view, add, edit, and delete dictionary terms.
When adding a term, users must specify the term (maximum), select a dialect, and optionally add up to 10 "sounds like" words to improve phonetic recognition. Users must include at least three example phrases (maximum 20 phrases) containing the term, and can configure a "Display As" field (50 character limit) to control how terms appear in transcripts. Dictionary entries apply to specific transcription engines—either Genesys Native Voice Transcription or Extended Voice Transcription Services (EVTS) using Amazon Web Services (AWS) Transcribe—and are tied to specific dialects.
The system enforces technical constraints: all entries are processed in lowercase, special characters and capitalization are not supported, and only standard alphanumeric characters (a–z, 0–9) are permitted. For EVTS dictionary entries specifically, numbers must be spelled out, hyphens and unsupported special characters are prohibited, acronyms require periods, and multi-word terms use spaces. Dictionary entries progress through status states—Pending, Active, Failed, or Invalid—indicating application success. Entries are applied automatically after saving, with updates typically taking effect within 15 minutes. Changes apply to new interactions, though ongoing interactions may not immediately reflect updates.
Access requires appropriate permissions including View, Add, Edit, and Delete permissions under Speech and Text Analytics > dictionaryterm. Users can edit existing terms to modify "sounds like" terms and example phrases, or delete terms entirely from the dictionary. Dictionary entries are engine-specific and not automatically migrated between transcription engines; users must manually recreate terms when switching engines, though identical terms can exist across multiple engines. The system validates entries during creation and editing, returning errors for unsupported characters, invalid formatting, or unsupported dialect-engine combinations