RFI Captioning and Transcription HM0476-22-I-CT.pdf

HM0476-22-I-CT

Captioning and Transcription

REQUEST FOR INFORMATION

Associate Director for Support Resources (DSR)

1.0 DESCRIPTION & PURPOSE

1.1 The National Geospatial-Intelligence Agency (NGA) in support of the Captioning and Transcription (C&T) program is seeking information on how an interested contractor could support C&T needs. This includes automated captioning of videos and producing written transcripts to support the Deaf and Hard of Hearing

Community (DHHC), Office of the Inspector General (OIG), and other agency stakeholders.

1.2 The purpose of this Request for Information (RFI) is to gain information on current industry capabilities that would meet NGA’s requirements specified in Appendix A.

1.3 THIS IS A REQUEST FOR INFORMATION (RFI) ONLY. This RFI is issued solely for information and planning purposes – it does not constitute a Request for Proposal (RFP) or a promise to issue an RFP in the future. This RFI does not commit the Government to contract for any supply or service whatsoever. Further, NGA is not at this time seeking proposals and will not accept unsolicited proposals.

Responders are advised that the U.S. Government will not pay for any information or administrative costs incurred in response to this RFI. All costs associated with responding to this RFI will be solely at the interested party’s expense. Not responding to this RFI does not preclude participation in any future RFP, if any is issued.

2.0 BACKGROUND

NGA has a mission-critical need for a robust video captioning capability to support the deaf or hard of hearing workforce. The desired automated solution will apply captions to the videos in real-time or allow uploading of an existing video for caption processing.

Additionally, NGA needs an automated solution that processes the audio from OIG meetings to provide timely and accurate written transcripts.

2.1 PLANNED ACQUISITION: At this stage, this is only an RFI to industry to learn what is currently in development or in the marketplace to support the requirements outlined in Appendix A.

2.2 PERFORMANCE:

2.2.1 Location(s): Springfield, VA and/or St Louis, MO.

2.3 SECURITY REQUIREMENTS : The following security requirements are anticipated to apply to this acquisition:

2.3.1 Contractors are required to have active Top Secret/Sensitive

Compartmented Information (TS/SCI) clearances for access to NGA facilities, when performing duties within classified environments, and for access to classified NGA computer systems.

2.3.2 Captioning and transcription processing must be performed in government approved unclassified and classified domain environments.

2.4 ORGANIZATIONAL CONFLICT OF INTEREST: In accordance with FAR Part 9.5, please discuss any performance which may lead to an OCI and how the Offeror would mitigate, avoid, or neutralize the conflict.

3.0 REQUESTED INFORMATION

Responses must only contain UNCLASSIFIED information and be marked

“UNCLASSIFIED” on all pages in both the header and footer. No classified information may be included in your response.

Proprietary and classified concepts may be included in the submittal. Input on the information contained in the responses may be solicited and reviewed by NGA non- Government consultants or experts who are bound by appropriate non-disclosure agreements.

3.1 ADMINISTRATIVE

Information to include the following as a minimum:

3.1.1 Name, mailing address, overnight delivery address (if different from mailing address), phone number, fax number, company website, and e-mail of designated point(s) of contact.

3.1.2 Business Type.

Based upon North American Industry Classification System (NAICS) code 512191 Captioning, the responder is to provide answers to the following questions:

To be considered for a Small Business set-aside, at least 50 % of the cost of contract performance incurred for personnel shall be expended for employees of the responder. (FAR 52.219-14, Limitation of Subcontracting)

RESPONSE

Are you a Small Business? (FAR 19.102), If so, which type: ☐ YES ☐ NO

Small Disadvantaged small business? (FAR 19.304) ☐ YES ☐ NO

Service Disabled small business? (FAR 19.14) ☐ YES ☐ NO

8(a) Small Disadvantaged small business? (FAR 19.8) ☐ YES ☐ NO

HUBZone small business? (FAR 19.13) ☐ YES ☐ NO

Woman-Owned Small Business? (FAR 19.15) ☐ YES ☐ NO

Involved in a mentor and/or protégé program? (DFARS 219.71) ☐ YES ☐ NO

3.1.3 Recommendation for alternative NAICS Code? Provide rationale for any NAICS Code applicable to the planned acquisition other than the NAICS code identified above.

3.1.4 DUNS Number: _________________

3.1.5 Commercial and Government Entity (CAGE) Code: _______________ http://www.census.gov/cgi-bin/sssd/naics/naicsrch?chart=2012 http://www.naics.com/naicsfiles/Size_Standards_Table2013.pdf http://www.acquisition.gov/far/current/html/Subpart%2019_1.html http://www.acquisition.gov/far/current/html/Subpart%2019_3.html http://www.acquisition.gov/far/current/html/Subpart%2019_14.html#wp1093836 http://www.acquisition.gov/far/current/html/Subpart%2019_8.html#wp1092796 http://www.acquisition.gov/far/current/html/Subpart%2019_13.html#wp1094801 http://www.acquisition.gov/far/current/html/Subpart%2019_15.html#wp1094868 http://www.acq.osd.mil/dpap/dars/dfars/html/current/219_71.htm

3.1.6 Defense facility security clearance? What type? What level?

3.1.7 Accounting system, if applicable? Date of last audit? Audit performed by? Determined adequate/inadequate? __________________________

3.1.8 Purchasing system? Date determined adequate or inadequate? ________

3.1.9 Timekeeping system? Date of last audit? Who performed? Determined adequate/inadequate? ______________________________________

3.1.10 Do you hold a GSA Federal Supply Schedule (FSS) contract applicable to the NAICS code (s) identified above? If yes, provide contract number(s). _

3.1.11 Are you registered in SAM? __________________________________

3.1.12 Are you interested in: Prime contract? Teaming to include Joint Venture?

Subcontractor opportunities? _________________________________

3.1.13 Include additional details that are not already requested based on the constraints of the information request.

3.2 EXPERIENCE

3.2.1 COMMERCIAL.

Provide relevant details on the responder providing the same or similar services offered or made to the general public or to non-governmental entities for purposes other than governmental purposes in the last three (3) years. Relevant details to NGA’s RFI should include, but not be limited to, information regarding the contract value, size and length of the effort, responder performing as a prime or subcontractor, customary practices (warranty, financing, discounts, contract types, etc.) under which the sales of the service(s) are made, security details, the customary practices regarding customizing, modifying, or tailoring of a service(s) to meet customer needs and associated costs, the kinds of factors that are used to evaluate performance; the kinds of performance incentives used; the kinds of performance assessment methods commonly used; the common qualifications of the people performing the services; and requirements of law and/or regulations unique to these service(s), that can demonstrate the responder’s abilities and capacity to meet NGA’s stated requirements.

3.2.2 GOVERNMENT.

Provide relevant details on the responder providing the same or similar services offered or made to Government agencies to include DOD in the last three (3) years. Relevant details to NGA’s potential acquisition should include, but not be limited to contract value, information regarding the contract number, agency, responder performing as a prime or subcontractor, size and length of the effort, type of pricing and/or cost, security details, the kinds of factors used to evaluate performance; the kinds of performance incentives; the kinds of performance assessment methods; the qualifications of the people performing the services; and any unique terms and conditions, that can demonstrate the responder’s abilities and capacity to meet NGA’s statement of objectives.

3.3 CAPABILITIES

The following is the responder’s capabilities to meet the proposed solution requirements set forth in Appendix A.

3.3.1 Government requirements needed to support the RFI statement of requirements

3.3.1.1 Requirement capabilities that the responder currently possesses.

Responses should include the relevant information regarding specific skills, experience, and security clearances that its employees, by labor category, currently possess in performing these services, and any needed hardware or software.

3.3.1.2 Detail the service capabilities that the responder currently does not possess in order to meet the list of validated requirements

3.3.2 Recommended type of pricing for this type of service (e.g. FFP, FPIF CPIF)

3.3.3 Extent to which the responder has the ability to meet the proposed acquisition

3.3.4 Notional schedule and type of plan for transition-in and transition-out (based on previous or similar work efforts).

3.3.5 Security capabilities and plans that demonstrate the ability to meet NGA’s security requirements beginning at contract award and throughout the

POP.

3.3.6 Extent to which the responder is aware of any potential OCI issues in accordance with FAR Part 9.5 or any non-mitigatable OCI related to current development work at NGA.

3.4 RECOMMENDATIONS

The responder is invited to provide information and recommendations for fashioning this proposed acquisition. Recommendation areas may include the type of contract, anticipated contract terms & conditions, incentives, variations in delivery schedule, price and/or cost proposal support, and data requirements, contract pricing, and any other areas that the responder believes is relevant for the Government to achieve its stated objectives. The responder may opine regarding what portion of the work could be set-aside for small businesses if all the required work cannot be provided by a small business. Details to support any recommendations for partial small business set-asides should be included.

4.0 RESPONSES

4.1 Interested parties are directed to respond to this RFI with a white paper.

4.2 White paper shall be in Microsoft Word for Office or compatible format and shall not exceed 15 pages. A page is defined as each face of an 8½” x 11” sheet with information contained within a one-inch margin on all sides. Font type shall be Times Roman 12 point.

4.2.1 Interested parties are encouraged to utilize the Comment Matrix format (Appendix B) in their white paper when providing comments, questions, and/or suggestions to the RFI.

4.3 Responses containing the white paper are due no later than 5:00 pm ET on 17 Nov

2022. Responses shall be limited to and submitted via e-mail only as a message attachment to eliana.c.york@nga.mil with the message subject line “Response to

NGA RFI HM0476-22-I-CT”.

4.4 Proprietary information, if any, should be minimized and MUST BE CLEARLY MARKED. To aid the Government, please segregate proprietary information. Please be advised that all submissions become Government property and will not be returned.

5.0 MEETINGS AND DISCUSSIONS

The Government representatives may or may not choose to meet with potential RFI service provider. Such meetings and discussions would only be intended to get further clarification of potential capability, especially any development and certification risks.

6.0 QUESTIONS

Questions and the whitepaper regarding this RFI should be submitted in writing by e-mail to eliana.c.york@nga.mil. Questions should be submitted no later than 5:00pm 7 Nov 2022.

7.0 SUMMARY

This RFI is only intended for the Government to identify sources that can provide the needed services to fulfill the Government’s statement of requirements. The information provided in this RFI is subject to change and is not binding to the Government. The Government has not made a commitment to procure any of the RFI requirements discussed, and release of this RFI should not be construed as such a commitment or as authorization to incur cost for which reimbursement would be required or sought. All submissions become Government property and will not be returned.

mailto:eliana.c.york@nga.mil mailto:eliana.c.york@nga.mil

APPENDIX A: LIST OF VALIDATED REQUIREMENTS FOR

CAPTIONING AND TRANSCRIPTION CAPABILITY NEED

Category Reqt # Validated Requirement

Audio to Text (Transcription Process)

001 Candidate systems shall provide capability of integration with voice messaging systems and production of voice message transcripts.

Automated Speech Recognition System (ASR)

002 Candidate system shall provide the ability to upload audio tracks and video with audio track data and generate transcripts of the data.

003 Candidate system shall provide the ability to upload NGA-specific development data consisting of audio track and video with audio track data files to their system.

004 Candidate system shall have an interface which permits other systems to access transcripts and captions both in real-time (streaming captions and/or transcripts for live display) and post-event (captions and transcripts produced as a single file).

005 Candidate system on secure domains shall execute capabilities wholly within the domain in which it is installed. Candidate system operating in secure domains shall not reach back to corporate servers for any reason, including license verification, authentication, model training, lexicon enhancement, speech to text (STT) execution.

006 Candidate system shall be cloud compatible to comply with NGA architecture standards and to provide cost and processing efficiency.

007 Candidate system shall be able to be implemented as a web service.

008 Candidate systems for implementation on unclassified domains may utilize vendor (corporate) servers for capability execution, including license verification, authentication, model training, lexicon enhancement, speech to text (STT) execution. Provided applicable security policies are observed.

010 Candidate system shall be capable of uploading files of completed event recordings to support post-event captioning.

Candidate system shall permit both audio and video with audio track file ingest.

011 Candidate system shall have the ability to ingest live or real-time audio and video and production of captions and video real-time.

Candidate system shall have the ability to ingest a variety of streaming media protocols for maximum flexibility.

012 Candidate system shall have the ability to realign corrected transcript with audio through forced alignment.

Customization 013 Candidate system shall possess a virtual meeting platform capability and shall provide the ability to generate automated captions and transcripts of meetings, export recorded video and automated meeting transcript provided by an automated speech to text tool for correction in a transcript editing system, then provide the ability to associate the corrected transcript with the original meeting recording in place of the automated captions or transcript.

014 Candidate system shall provide the ability to detect songs present in audio and video and transcribe song lyrics. Lyrics shall be annotated with musical note or comment like (music) to indicate these are lyrics.

015 Candidate system shall accommodate persistent user profiles that retain user preferences for customization of caption display formatting, including characteristics identified in Requirements # 67 (display of edited captions) 68 (display of raw ASR captions).

016 Candidate system shall enable transcript editor to choose to correct a customized percentage of the transcript with the highest error.

017 Candidate system shall possess an integrated system transcript editing tool.

Candidate system shall include the ability to add a label to the transcript and caption file indicating the percentage of the transcript which has been reviewed.

Mobile Captioning 018 Candidate system shall provide real-time speech to text transcription on mobile devices compatible with NGA security requirements which can be used on an ad hoc basis for 1:1 conversations, events/meetings without interpreters, etc. Mobile devices cannot connect to wi-fi, Bluetooth, or any wireless network.

May be connected to NGA networks via USB only to download transcripts.

019 Provide real-time speech to text transcription on mobile devices compatible with NGA security requirements which can be used on an ad hoc basis for 1:1 conversations, events/meetings without interpreters, etc. Mobile devices can connect to wi-fi, Bluetooth, or wireless network (Select location).

May also be connected to NGA networks via USB only to download transcripts.

Models 020 Candidate system shall permit development of customized models for differing acoustic environments and use cases.

021 Candidate system shall allow for selection of appropriate model for environment and use case to maximize accuracy and performance.

022 Candidate system shall have the ability to tune the acoustic models on CLASSIFIED audio recordings.

Candidate system shall have the ability to conduct training of the system internal to the CLASSIFIED domain and not reachback to Corporate server or cloud services.

023 Candidate system shall provide the ability to tune the acoustic model on UNCLASSIFIED audio recordings.

For system implemented on UNCLASSIFIED domains, this training may occur using vendor servers, external to NGA networks, if applicable security policies are observed.

024 Candidate system shall provide capability of continuous training of CLASSIFIED Language model through ingest of corrected transcripts during re-alignment processing.

Candidate system shall have continuous/iterative updating that occurs wholly within a CLASSIFIED domain.

025 Candidate system shall provide the ability of training CLASSIFIED Language models with training data.

This training shall occur wholly within the CLASSIFIED network for system implemented on secure domains.

026 Candidate system shall provide the ability for continuous, iterative training of UNCLASSIFIED Language model through ingest of corrected transcripts during re-alignment processing.

027 Candidate system shall provide the ability for training UNCLASSIFIED Language model with training data.

028 Candidate system shall provide capability for continuous, iterative training of CLASSIFIED Lexicon through ingest of corrected transcripts during re-alignment processing. This training shall occur wholly within the CLASSIFIED network for system implemented on secure domains.

029 Candidate system shall provide capability for training CLASSIFIED Lexicon through use of training data.

This training shall occur wholly within the CLASSIFIED network for system implemented on secure domains.

030 Candidate system shall provide capability for continuous, iterative training of UNCLASSIFIED Lexicon through ingest of corrected transcripts during re-alignment processing.

031 Candidate system shall provide capability of training UNCLASSIFIED Lexicon through use of training data.

032 Candidate system shall provide ability to export model pack from Automated Speech Recognition (ASR) system for use in a different ASR tool.

033 Candidate system shall provide ability to export model pack from Automated Speech Recognition (ASR) system for use in separate instance of same ASR tool.

034 Candidate system shall provide ability to import a model pack from the Automated Speech Recognition

(ASR) module of a separate instance of same ASR tool.

035 The system shall be able to store multiple models for use in varying sound environments and use cases.

036 The system shall allow import of a pre-trained KALDI model pack.

037 Candidate system shall provide the details of the training and test data upon which their system is trained out of the box.

Output - Captions 038 Candidate system shall provide streaming, real-time captions for live display during events when provided with a streaming audio feed. The system shall allow for selection from multiple caption file output formats and customization of the display formatting of the captions for maximum flexibility and interoperability with supporting system, such as streaming media servers.

039 Candidate system shall provide captions as a caption format text file on demand upon completion of captioning an event, whether live captioned or post-event captioned. The system shall allow for selection from multiple caption file output formats and customization of the display formatting of the captions for maximum flexibility and interoperability with supporting system, such as streaming media servers.

Output - Transcripts 040 Candidate system shall allow for customization of transcript output, including presence/absence of time stamps, Header information (for ex. Date/time generated, audio/video filename, Manual input of info, such as meeting/event name + date, Classification, line numbering, etc.) to ensure for maximum interoperability and end use flexibility.

041 Candidate system shall provide transcripts as a customizable text file on demand upon completion of transcription of an event, whether live transcribed or post-event transcribed. The system shall allow for customization of the transcript file as noted in Requirement # 040 for maximum flexibility and interoperability with supporting system, such as streaming media servers and transcript correction for high accuracy caption production.

042 Candidate system shall provide streaming, real-time transcription for live display during events when provided with a streaming audio feed. The system shall allow for selection from multiple transcript file output formats and customization of the display formatting of the transcript as noted in Requirements #040, 067, and 068 for maximum flexibility and interoperability with supporting system, such as streaming media servers and low-latency transcript correction for production of high accuracy transcripts.

043 Candidate system shall be capable of providing word level confidence scores which can be accessed by

NGA engineers to facilitate error identification in transcripts in support of transcript correction.

Performance 044 Candidate system shall provide sentence capitalization and capitalization of proper names of within caption and transcript output with a high level of accuracy. Candidate system shall provide the level of accuracy that the system can attain for this capability.

045 Candidate system shall provide punctuation of caption and transcript output with a high level of accuracy, including commas, periods, quotation marks, and question marks. Candidate system shall provide the level of accuracy that the system can attain for this capability.

046 Candidate system shall identify sentences in the caption and transcript output with a high level of accuracy. Candidate system shall provide the level of accuracy that the system can attain for this capability.

047 Following training with NGA development data, candidate system shall demonstrate a transcript word error rate on NGA test data without correction of 25% or lower. This word error rate will be determined with the SCLite tool kit.

048 Candidate system shall demonstrate offline recognition rate at a minimum of 10x faster than real time (1 hour of audio shall be transcribed in a maximum of 6 minutes)

Post-Processing 049 Candidate system must produce transcripts and captions which include punctuation.

050 Candidate system must produce transcripts and captions which include sentence capitalization.

Pre-Processing 051 Candidate system shall produce transcripts and captions which are diarized, identifying segments spoken by different speakers.

052 Candidate system shall provide speaker id of transcript and captions, identifying segments spoken by each unique speaker’s in audio track.

z_Media Server 053 Candidate system shall provide streaming media server capabilities on CLASSIFIED domain, permitting streaming of multiple simultaneous video and audio files on demand.

054 Candidate system shall provide streaming media server capabilities on UNCLASSIFIED domains permitting streaming of multiple simultaneous video and audio file on demand z_Storage and Retrieval 055 Candidate system shall provide storage system for captioned video and transcribed audio files.

056 Candidate system shall provide capability to stream captioned video and transcribed audio from storage system, so that audio and video does not need to be downloaded to local system to play.

057 Candidate system shall provide retrieval capability for captioned video and transcribed audio using metadata associated with files.

058 Candidate system shall provide retrieval capability for captioned video and transcribed audio at time stamp level using transcripts of video and audio files.

059 Candidate system shall provide storage of video and audio transcripts, linked to captioned video and transcribed audio files.

z_Transcript Editing 060 Candidate system shall provide a text editing interface for real-time CLASSIFIED transcript correction simultaneous with live caption production. Candidate system shall execute capabilities wholly within the domain in which it is installed. Cannot reach back to corporate servers for any reason, including license verification, authentication, model training, lexicon enhancement, speech to text (STT) execution.

061 Candidate system shall provide a text editing interface for real-time UNCLASSIFIED transcript correction simultaneous with live caption production.

062 Candidate system shall provide a text editing interface for CLASSIFIED transcript correction with integrated audio playback. Candidate system shall execute capabilities wholly within the domain in which it is installed. Cannot reach back to corporate servers for any reason, including license verification, authentication, model training, lexicon enhancement, speech to text (STT) execution.

063 Candidate system shall provide a text editing interface for UNCLASSIFIED transcript correction with integrated audio playback.

z_Video Editing 064 Candidate system shall provide a video editing service or software supporting advanced formatting and customization of open and closed captions z_Video Production 065 Candidate system shall produce open and closed-captioned video using corrected transcript.

066 Candidate system shall produce open and closed-captioned video with captions produced by ASR system.

067 Candidate system shall support display customization of open and closed-captioned video using corrected transcript.

The candidate system shall provide system capabilities for the following attributes for customization:

-Background color -Strength of contrast -Font size -Font color -Font type -Placement of captioning/location on the display -Number of lines to display -Characters per line -% of Opacity of background color

068 Candidate system shall support display customization of open and closed-captioned video with captions produced by ASR system.

The candidate system shall provide system capabilities for the following attributes for customization:

-Background color -Strength of contrast -Font size -Font color -Font type -Placement of captioning/location on the display -Number of lines to display -Characters per line -% of Opacity of background color

APPENDIX B: COMMENT MATRIX FOR THE CAPTIONING AND TRANSCRIPTION CAPABILITY NEED

REQUEST FOR INFORMATION (RFI)

COMMENT # PAGE # SECTION COMMENT/QUESTION/SUGGESTION