Voice Dictionary Entries
To ensure AI agents correctly recognize and transcribe specific pronunciations—for example, unique product names, abbreviations, and regional or company-specific phrases. These entries can be applied across the Vault CRM instance or applied only to specific user languages and countries.
Veeva's AI agents natively transcribe standard medical terms and common proper nouns without requiring a specialized medical dictionary; however, abbreviations, unique brand names, and abbreviations-as-proper-nouns are not automatically recognized and must be added as a Voice Dictionary Entry to ensure accurate transcription.
Who can use this feature?
- Business Admin Users - Browser
- Creating Voice Dictionary Entries
- End Users - iPad, iPhone
- Using Custom Pronunciations
- A Veeva AI license is required
Configuring Voice Dictionary Entries for
To configure this feature:
-
Ensure Configuring Vault AI in CRM is complete.
-
Navigate to Admin > Users & Groups > Permission Sets.
-
Select the appropriate permission set.
-
Select the Objects tab.
-
Grant the following permissions:
Object
Object Permission
Object Types
Fields
Field Permission
voice_dictionary_entry__v
CRED
core_object_type__v
- country__v
- language__v
- name__v
- phrase__v
- pronunciation__v
Edit
To configure this feature:
-
Ensure Configuring Vault AI in CRM is complete.
-
Navigate to Admin > Users & Groups > Permission Sets.
-
Select the appropriate permission set.
-
Select the Objects tab.
-
Grant the following permissions:
Object
Object Permission
Object Types
Fields
Field Permission
voice_dictionary_entry__v
R
core_object_type__v
All Fields
Read
-
Navigate to Business Admin > Objects > vmobile_object_configuration__v.
-
Ensure voice_dictionary_entry__v VMOCs are active for the appropriate platforms.
Creating Voice Dictionary Entries as
To create Voice Dictionary Entries for custom pronunciations:
-
Navigate to Business Admin > Objects > voice_dictionary_entry__v.
-
Select Create.
-
Populate the Phrase field with the phrase or word’s written spelling. For example, Natevba.
-
Populate the Pronunciation field with the phrase or word’s pronunciation, using the X-SAMPA Transcription Scheme.
-
Select a Language and/or Country for the entry (optional):
- Languages are matched to the language__sys field on the user’s user__sys record. If a language is not selected on the Voice Dictionary Entry, the entry applies to all user languages.
- Countries are matched to the country__sys field on the user__sys record. If a country is not selected on the Voice Dictionary Entry, the entry applies to all user countries.
- If both the Language and the Country fields are left blank on a Voice Dictionary Entry, the entry applies to all languages and countries
-
Select Save.
Each entry includes:
- Phrase – The written spelling of the word or phrase (for example, Natevba)
- Pronunciation – The word’s pronunciation, written using the X-SAMPA Transcription Scheme, a standardized way to represent speech sounds using plain text characters
X-SAMPA (Extended Speech Assessment Methods Phonetic Alphabet) provides a device-independent way to specify pronunciations. It ensures consistency across regions and languages, even when device or platform-level phoneme support differs.
For example, in an organization with product names like Cholecap, Restolar, and Natevba, an admin would populate the Phrase and Pronunciation on a Voice Dictionary Entry as follows:
|
Phrase |
Pronunciation (X-SAMPA Transcription) |
|---|---|
|
Cholecap |
k"ol@k{p |
|
Restolar |
r"Est%olAr |
|
Natevba |
n"OtEvbO |
Because iOS phoneme support varies by region and device, some X-SAMPA symbols may not be supported everywhere. Vault CRM does not validate phoneme compatibility, so unsupported symbols may result in incorrect or unrecognized pronunciations.
Veeva recommends testing the Voice Dictionary Entries with end users who reflect the full range of accents, dialects, and languages in use before making agentic AI dictation capabilities, available to all users.
Using Custom Pronunciations as
When users’ speech matches the Pronunciation value on a Voice Dictionary Entry, the AI agent uses the specified Phrase text to accurately transcribe the sound. Voice Dictionary Entries apply for the following AI functionalities:

