Automated Speech Recognition

UN Secretariat
Automated Speech Recognition Request for information

Reference: RFIUNPD21995
Beneficiary countries or territories: Multiple destinations (see 'Countries' tab below)
Registration level: Basic
Published on: 18-Oct-2023
Deadline on: 16-Nov-2023 23:59 (GMT -4.00)

Description
The United Nations is seeking information from potential vendors to understand the capabilities and solutions available on the market for an Automated speech recognition (ASRs), transcription and translation, especially in the context of peacekeeping settings. The organization is interested in identifying solutions for automated transcription and translation of audio streams of various quality, from online and offline sources, in both official and unofficial languages spoken - including regional dialects – in countries of UN operations. The purpose of this RFI is to gather details about your product offerings and capabilities in this domain, mainly to: 1. Identify the state-of-the-art technologies and solutions for speech recognition and translation in multiple languages and dialects 2. Evaluate the accuracy, speed, scalability and security of the available solutions 3. Assess the feasibility and cost-effectiveness of implementing such solutions for the UN's operational needs 4. Explore the potential benefits and challenges of using such solutions for enhancing capabilities of UN applicatoins I. Background and Objectives: The UN is responsible for relief, peacekeeping and monitoring operations in various regions around the world. To effectivelly implement their mandate, UN entities rely on information management, by using different tools and methods to gather, analyze and understand audio data from various sources and in various quality. However, there is limited availability of transcription and translation solutions for certain local dialects, or even whole languages, especially if concerning remote populations. Therefore, we are looking for a comprehensive solution, or components of a solution, that can: 1. Provide ASRs (speech-to-text) for languagues spoken primarily in: Democratic Republic of Congo, Central African Republic, Yemen, Lebanon, Somalia, Sudan. The organisation is interested also in solutions for local dialets for official languages (for example Somali Arabic, Yemen Arabic), also any local language(s) for countries above. Full list of countries of interest, where UN missions operate can be found here: a. Where we operate | United Nations Peacekeeping b. DPPA Around the World | Department of Political and Peacebuilding Affairs. 2. Provide automated translation (text-to-text) solutions to English and French from languages listed above. 3. Tools that could help the UN to rapidly and efficiently develop and train new ASRs for languages currently unavailable on the market  4. The delivered ASR and translation solutions should be compatible for deployment to either AWS transcribe or Azure (either within those environments, which is preffered, or via API).       

Email address: whitei@un.org
Isaam White