BEGIN:VCALENDAR METHOD:PUBLISH PRODID:Microsoft Exchange Server 2010 VERSION:2.0 BEGIN:VTIMEZONE TZID:tzone://Microsoft/Utc BEGIN:STANDARD DTSTART:16010101T000000 TZOFFSETFROM:+0000 TZOFFSETTO:+0000 END:STANDARD BEGIN:DAYLIGHT DTSTART:16010101T000000 TZOFFSETFROM:+0000 TZOFFSETTO:+0000 END:DAYLIGHT END:VTIMEZONE BEGIN:VEVENT ORGANIZER;CN=Millicent Ochieng:mailto:mochieng@microsoft.com ATTENDEE;ROLE=REQ-PARTICIPANT;PARTSTAT=NEEDS-ACTION;RSVP=FALSE;CN=Millicent Ochieng:mailto:mochieng@microsoft.com ATTENDEE;ROLE=OPT-PARTICIPANT;PARTSTAT=NEEDS-ACTION;RSVP=FALSE;CN=Stephanie Nyairo:mailto:snyairo@microsoft.com DESCRIPTION:\n\n[cid:image001.png@01DD5FC7.26EB5330]\n\n[cid:image002.png@0 1DD5FC7.26EB5330]\n\n\n\nHello\,\n\n\n\nJoin us for next month’s seminar as we explore how model architecture\, language typology\, and adaptation strategies influence speech recognition performance across African langua ges. We’re excited to host Mercy Muchai\, Research Engineer at Microsoft Research Africa.\n\n\n\nTalk: Paza: How Architecture\, Language Typology\ , and Adaptation Shape Speech Recognition in African Languages\n\nDate: Th ursday\, November 5\, 2026\nTime: 3:00 PM – 4:00 PM EAT\nLocation: MAGAD I\, 9th floor & Online\n\n\n\nAbstract:\n\nDespite rapid and meaningful ad vances in speech recognition\, building robust ASR systems for low-resourc e languages remains challenging. Key questions persist: how much data shou ld be collected and in what forms\, which model architectures offer the be st opportunities for training and deployment\, and what adaptation strateg ies are most effective for new domains and languages?\n\nIn this talk\, Mu chai will present an architecture-aware evaluation of 53 speech models and their accuracy-efficiency trade-offs across 39 African languages. Muchai will also discuss research findings from a systematic study fine-tuning th ree models from distinct architectural paradigms\, revealing how model arc hitecture\, language typology\, and adaptation strategies shape speech rec ognition performance and real-world deployment outcomes. This talk draws f rom research on the Paza Speech Playbook and recent publications.\n\nAbout the Speaker:\n\nMercy Muchai is a Research Engineer at Microsoft Research Africa in Nairobi. She leads the end-to-end AI/ML engineering efforts wit hin the lab. Her work focuses on building globally equitable AI tools and models that can reason over multimodal data\, with a particular emphasis o n improving the linguistic and cultural understanding of Generative AI for underrepresented languages and communities.\n\nHer recent work includes l eading the research and development of automatic speech recognition models for African languages\; including PazaBench\, the Paza Speech Models and co-creating the Gecko Playbooks - a collection of toolkits for building mo re equitable AI systems. Muchai is also a Computer Science graduate studen t at the University of Texas at Austin. She is passionate about building i nclusive technologies for the Global Majority and conducting research that bridges existing gaps through community-participatory approaches.\n\nYou are receiving this email because you have subscribed to the Microsoft Rese arch Africa\, Nairobi mailing list.\nUnsubscribe | Privacy Statement\n\n\n\n\n\n____________ ____________________________________________________________________\nMicr osoft Teams meeting\nJoin: https://teams.microsoft.com/meet/24771791892502 70?p=tqqc600jCi8LVuH68h\nMeeting I D: 247 717 918 925 027 0\nPasscode: 7vR2Px9M\n____________________________ ____\nNeed help? | System reference\nDial in by phone\n+1 323-849-4874\,\,238825926# Uni ted States\, Los Angeles\n0800 720 662\,\,238825926# Kenya (Toll-free)\nFind a local number\nPhone conference ID: 238 825 926#\nFor organizers: Meeting options | Reset dial-in PIN\n_____________________________ ___________________________________________________\n\n UID:040000008200E00074C5B7101A82E0080000000070BDD5B5C65FDD01000000000000000 0100000001741FB2D2D17614B8F92551A14B2D0CC SUMMARY:[MSR Africa & AI4G Seminars] Paza: How Architecture\, Language Typo logy\, and Adaptation Shape Speech Recognition in African Languages DTSTART:20261105T120000Z DTEND:20261105T130000Z CLASS:PUBLIC PRIORITY:5 DTSTAMP:20261019T100918Z TRANSP:OPAQUE STATUS:TENTATIVE SEQUENCE:0 LOCATION:Microsoft Teams Meeting\; Conf Room NAIROBI-WAIYAKI WAY/9.2A (7) MAGADI X-MICROSOFT-CDO-BUSYSTATUS:TENTATIVE X-MICROSOFT-CDO-INTENDEDSTATUS:BUSY X-MICROSOFT-CDO-ALLDAYEVENT:FALSE X-MICROSOFT-CDO-IMPORTANCE:1 X-MICROSOFT-CDO-INSTTYPE:0 X-MICROSOFT-SKYPETEAMSMEETINGURL:https://teams.microsoft.com/l/meetup-join/ 19%3ameeting_NGFhY2UyNWQtZGYxMy00YWJmLThjYWYtZDdjYTk5MTE4YWQx%40thread.v2/ 0?context=%7b%22Tid%22%3a%2272f988bf-86f1-41af-91ab-2d7cd011db47%22%2c%22O id%22%3a%2209d7c323-131f-4a12-a4be-f215c1e46936%22%7d X-MICROSOFT-SCHEDULINGSERVICEUPDATEURL:https://api-fd.scheduler.teams.micro soft.com/teams/72f988bf-86f1-41af-91ab-2d7cd011db47/09d7c323-131f-4a12-a4b e-f215c1e46936/19_meeting_NGFhY2UyNWQtZGYxMy00YWJmLThjYWYtZDdjYTk5MTE4YWQx @thread.v2/0 X-MICROSOFT-SKYPETEAMSPROPERTIES:{"cid":"19:meeting_NGFhY2UyNWQtZGYxMy00YWJ mLThjYWYtZDdjYTk5MTE4YWQx@thread.v2"\,"private":true\,"type":0\,"mid":0\," rid":0\,"uid":null} X-MICROSOFT-ONLINEMEETINGINFORMATION:{"OnlineMeetingChannelId":null\,"Onlin eMeetingProvider":3} X-MICROSOFT-ONLINEMEETINGTOLLFREENUMBERS:+254 800 720 662 X-MICROSOFT-DONOTFORWARDMEETING:FALSE X-MICROSOFT-DISALLOW-COUNTER:TRUE X-MICROSOFT-LOCATIONS:[ { "DisplayName" : "Microsoft Teams Meeting"\, "Loca tionAnnotation" : ""\, "LocationSource" : 0\, "Unresolved" : false\, "Loca tionUri" : "" }\, { "DisplayName" : "Conf Room NAIROBI-WAIYAKI WAY/9.2A (7 ) MAGADI"\, "LocationAnnotation" : ""\, "LocationSource" : 5\, "Unresolve d" : false\, "LocationUri" : "cfww92A@microsoft.com" } ] BEGIN:VALARM DESCRIPTION:REMINDER TRIGGER;RELATED=START:-PT15M ACTION:DISPLAY END:VALARM END:VEVENT END:VCALENDAR