Define the purpose
The institution and CareCompile would specify meeting or dictation cases, limits on use, and the questions the evaluation must answer.
CareCompile speech program
A counterpart for Salvadoran voices and texts
The main work ahead is not selling a product. It is building an institutional relationship that can define what data is needed, who may contribute it, and how it is governed. No regional audio has been collected for this program.
In development for internal use. Our own model is not trained yet.
01 / VOCERO
The institutional request
CareCompile seeks a first counterpart in El Salvador to open the conversation before any material is recorded or transferred.
Review consent and custody
The conversation should cover consent language, withdrawal, deletion, access, storage, and custody of the test set.
Contribute local knowledge
Institutions, universities, and media organizations can guide vocabulary, names, registers, and regional variation, as well as linguistic review.
02 / VOCERO
Data anticipated in the plan
The quantities and collections described are planning assumptions. They do not represent data already received.
Voices with transcripts
The plan covers scripted reading and spontaneous speech from consenting adults, with representation across regions, ages, and genders.
Dictation clips
Original scripts would cover commands, numbers, dates, and regional names with the same explicit consent for training.
Texts and meetings
Regional texts require a recorded license. Meeting-style examples would be synthetic; a real meeting could be used only with consent.
03 / VOCERO
Material the program does not accept
The published rules exclude material whose origin, authorization, or sensitivity prevents responsible use.
Real patient data
The program does not accept real patient information.
Conversations without consent
Private conversations are not accepted unless every person involved has authorized their use.
Minors and material without rights
Recordings of minors and material the person or institution has no right to share are not accepted.
04 / VOCERO
Proposed governance
The program follows the eight published principles and controls that still need to be agreed and put into practice.
Limited use and withdrawal
Data would be used only to build and evaluate Vocero, would not be resold, and contributors could request withdrawal and deletion.
Separated test data
The design reserves a sealed test set with different speakers, held by a person who does not train models. The set does not yet exist.
Traceability
Every source would have a license note; the test manifest would use a SHA-256 fingerprint published before training.
Explore the program
Related pages
From a meeting to reviewable minutes
Explore Vocero's design for transcribing meetings and producing minutes with decisions, tasks, and source citations. In development; the demo is synthetic.
Explore the scopeDictation for regional Spanish
Vocero investigates dictation for Salvadoran and Central American Spanish. Try the implemented command rules and explore the regional speech plan.
Explore the scopeArchitecture and evaluation before claims
Review Vocero's proposed architecture, training plan, and Vocero-Bench design for measuring speech, dictation, and minutes by country. There are no results yet.
Explore the scopeEL SALVADOR / CARECOMPILE
Designate an institutional counterpart in El Salvador
We are looking for an institution willing to discuss regional meetings, voices, and texts. We explain purpose, consent, and governance before recording or receiving data.