Tenders Are Invited For Software Package And Information Systems – Purchase, Installation And Adjustment Of The Slovenian Speech Recognizer

Software Package And Information Systems – Purchase, Installation And Adjustment Of The Slovenian Speech Recognizer. The Subject Of The Public Contract Is The Purchase, Adaptation And Installation Of A Slovenian Speech Recognizer. The Details Of The Subject Of The Public Procurement Are Given In Chapter 3.3 Technical Requirements Of The Client, Where The Client Specifies Technical, Functional And Process Requirements That Arise From The Special Needs Of Subtitling Audiovisual Content At The Public Institution Radiotelevisija Slovenija. The Requirements Include The Purchase, Installation, Start-Up And Adjustment Of Server Software For Recognizing Slovenian Speech (Hereinafter Server). The Client Also Requires The Development, Installation And Commissioning Of Client Software For Deferred Transcribing And For Monitoring And Monitoring System Performance. The Purpose Of The Order Is The Establishment Of A Modern, Neural Network-Based E2e (English End-To-End) System With The Latest Architectures (E.G. Transformer Or Similar) For The Recognition Of Slovenian Speech, Which Will Provide Core Technical Functionality To The Work Processes For The Preparation Of Subtitles. Due To The Specific Requirements Of The Field And Compliance With Legal Requirements Regarding Quality, The Provider Must Offer A System That Can Be Adapted To Linguistic And Technical Requirements And Specialized (Specialization) With The Help Of The Clients Learning Data For At Least 3 Specific Areas Of Use: Informative, Cultural And Sports Content. In Doing So, The Client Will Prepare A Learning Corpus With At Least 100 Hours Of Audio Recordings With Manually Prepared And Verified Verbatim Reference Transcripts For Each Domain. The Server Software Must Be Designed Modularly, With Clearly Demarcated Components In Such A Way That The Gradual Expansion Of Functionality And Capacity And Easy Upgrades To New Versions Of Models And Software Are Enabled. All Recognition Processes Take Place On The Clients Infrastructure In Its Data Center, Unless Otherwise Expressly Stated In This Documentation Regarding The Award Of A Public Contract. The Offeror Must Attach To The Offer The Technical Specification Of The Offered Software Server And All Other Proofs, Certificates And Other Documentation Defined And Requested By The Client In The Technical Specifications Of The Equipment. The Subject Of The Public Procurement Includes: • Purchase Of A Server License, Meaning The Purchase Of A Permanent License For All Server Components Of The Software, Which Are Necessary For Continuous Implementation Of Real-Time Recognition Of Input Speech In Audiovisual Content And Transcribing Speech On Demand. • Installation Of The Server And Client Applications On The Clients Hardware And Customization, Which Includes The Installation Of The Server Along With The Initial And Later Specialized Models And The Client Applications On The Clients Hardware In Its Data Center. It Also Covers The Installation Of The System Software Required To Set Up Control Modules To Monitor The System Metrics Of The Recognizer. • Creation Of A Project For Implementation (Pzi) And Development Of Client Software For The Use Of A Speech-To-Eye Recognizer