MLMI 2004 -- Accessing Multimodal Meeting Data: Systems, Problems and Possibilities -- Browsing Recorded Meetings with Ferret -- Meeting Modelling in the Context of Multimodal Research -- Artificial Companions -- Zakim - A Multimodal Software System for Large-Scale Teleconferencing -- Towards Computer Understanding of Human Interactions -- Multistream Dynamic Bayesian Network for Meeting Segmentation -- Using Static Documents as Structured and Thematic Interfaces to Multimedia Meeting Archives -- An Integrated Framework for the Management of Video Collection -- The NITE XML Toolkit Meets the ICSI Meeting Corpus: Import, Annotation, and Browsing -- S-SEER: Selective Perception in a Multimodal Office Activity Recognition System -- Mapping from Speech to Images Using Continuous State Space Models -- An Online Algorithm for Hierarchical Phoneme Classification -- Towards Predicting Optimal Fusion Candidates: A Case Study on Biometric Authentication Tasks -- Mixture of SVMs for Face Class Modeling -- AV16.3: An Audio-Visual Corpus for Speaker Localization and Tracking -- The 2004 ICSI-SRI-UW Meeting Recognition System -- On the Adequacy of Baseform Pronunciations and Pronunciation Variants -- Tandem Connectionist Feature Extraction for Conversational Speech Recognition -- Long-Term Temporal Features for Conversational Speech Recognition -- Speaker Indexing in Audio Archives Using Gaussian Mixture Scoring Simulation -- Speech Transcription and Spoken Document Retrieval in Finnish -- A Mixed-Lingual Phonological Component Which Drives the Statistical Prosody Control of a Polyglot TTS Synthesis System -- Shallow Dialogue Processing Using Machine Learning Algorithms (or Not) -- ARCHIVUS: A System for Accessing the Content of Recorded Multimodal Meetings -- Piecing Together the Emotion Jigsaw -- EmotionAnalysis in Man-Machine Interaction Systems -- A Hierarchical System for Recognition, Tracking and Pose Estimation -- Automatic Pedestrian Tracking Using Discrete Choice Models and Image Correlation Techniques -- A Shape Based, Viewpoint Invariant Local Descriptor.