辞書管理を理解する
Dictionary management allows you to fine-tune transcription accuracy by adding organization-specific terms that the engine may not recognize by default.
With dictionary management, Genesys Cloud voice transcription can better recognize unique language subtleties, brand names, industry-specific terms, and internal acronyms across languages and dialects. You can also define correct pronunciation and provide “sounds like” hints to improve recognition.
Examples of terms
- Product or brand names: ProSecure XL, Louis Vuitton
- Acronyms: RMA (Return Merchandise Authorization), IRS
- Internal terminology: Tier Zero Support, QuickPay Plan
Term attributes you can configure
- Sounds like: Provides phonetic hints to guide how the term should be interpreted, improving recognition of brand names or uniquely spelled words.
- Boost value: Increases the priority of a term, helping the transcription engine favor it over similar-sounding alternatives. Boost values are only available via the API.
When to use dictionary management
- When you have a manageable list of industry-specific terms such as product names, acronyms, or branded phrases that are critical to your business
- When you notice consistent recognition errors in transcripts (for example, key terms being misheard, split incorrectly, or omitted)
- When launching new products, services, or campaigns with terminology not commonly recognized by standard transcription engines
主な特徴
- A single dictionary list is used across the entire organization
- Each term is associated with a single dialect
- Administrators can search terms and filter by dialect
- Each dialect can contain up to 500 terms
- Administrators can add up to 10 alternative spellings per term to aid phonetic recognition
- Up to 20 example phrases can be added per term to provide context
- Each term can have a maximum of 30 characters
- Terms are case-insensitive
- Boost values can only be updated via the API
- Updates to terms, examples, “sounds like,” or boost values may take up to 30 minutes to propagate across the organization
- Dictionary management improves transcription accuracy in both standard and low-latency modes
The following table contains additional examples:
期間 | フレーズ例 | のように聞こえる |
---|---|---|
神経系 |
彼は神経疾患を患っている |
ニューロロジカル |
健康管理 |
資格のある医療提供者 |
健康管理 |
プライミング |
ポンプの呼び水 |
プライミング |
IRS |
IRSの監査 |
アイアールエス |
アクメ |
私の好きなブランドはアクメです |
ニキビ |
ルイヴィトン |
好きなブランドはルイ・ヴィトンです |
ルイ・ヴィト |
For more information, see Work with dictionary management.