Tiko Africa is looking to source a partner to conduct the collection of voice samples at Tiko clinics, Tiko pharmacies and households in Kenya. The voice collected shall assist Tiko Africa to solve 3 needs for our business.

• User Identification
  • Creating the ability for users to be uniquely identified on the platform with their voice. This is used in conjunction with either a Tiko card (with a unique number) or a phone number. Their voice is used to authenticate that they are the user for the profile on the system
• Continuation:
  • Identify the same user is coming back repeatedly to get SRH services or products
• Fraud Prevention / Prevention of system gaming
  • Prevent users from accessing other users accounts (IE If they find a Tiko card that belongs to someone else).

The voice sample collected will be used to train software on Kenyan voices to improve the identification of these voices in the Tiko platform.

Objectives

To collect identifiable data and voice samples from participants speakers, each with an enrollment sample and 5+ different verification samples. (IE Each voice collection will be a 6 samples for one user at minimum)

• All participants will have been informed on what the voice collection is for
• All participants need to consent to the voice being captured
• Verification calls must be at least one day after enrolment.
• Voice samples need to be collated to identifiable information for that voice
• Data specification listed in technical requirements
• Voice sample specification listed in technical requirements
• Sample allocation per household, facility and location can be found in sample allocation tables

The ideal partner would have the software, tools and capability to collect voices as needed and be located in Kenya. If the partner doesn’t have the capability, tools or software to perform an end to end service, this should be stated clearly in the proposal what part of the RFP can be done.

Technical requirements

• All information collected should be collected in a manner that it is linked.

Collect Consent

• Acknowledgement of participant accepting to have their voice collected

Collect Identifiable information (Questionnaire)

• Age of participant
• Birth Gender
• Language collected

Collect System information(System)

• Location (Ward, County etc)
• GPS Location

Voice Sample

• File format – Waveform Audio File Format (.wav) & 8KHz
• Number of samples listed in Tables 1 to 4 below
• 1 enrolment voice with net speech of 20 sec (at minimum 30 sec long in total)
• 5 verification voice with net speech of 15 sec

(Net speech is defined as the part of the recording that has voice)