Scientific bases of human-machine communication by voice.

The scientific bases for human-machine communication by voice are in the fields of psychology, linguistics, acoustics, signal processing, computer science, and integrated circuit technology. The purpose of this paper is to highlight the basic scientific and technological issues in human-machine communication by voice and to point out areas of future research opportunity. The discussion is organized around the following major issues in implementing human-machine voice communication systems: (i) hardware/software implementation of the system, (ii) speech synthesis for voice output, (iii) speech recognition and understanding for voice input, and (iv) usability factors related to how humans interact with machines.

The role of voice input for human-machine communication.

Optimism is growing that the near future will witness rapid growth in human-computer interaction using voice. System prototypes have recently been built that demonstrate speaker-independent real-time speech recognition, and understanding of naturally spoken utterances with vocabularies of 1000 to 2000 words, and larger. Already, computer manufacturers are building speech recognition subsystems into their new product lines. However, before this technology can be broadly useful, a substantial knowledge base is needed about human spoken language and performance during computer-based spoken interaction. This paper reviews application areas in which spoken interaction can play a significant role, assesses potential benefits of spoken interaction with machines, and compares voice with other modalities of human-computer interaction. It also discusses information that will be needed to build a firm empirical foundation for the design of future spoken and multimodal interfaces. Finally, it argues for a more systematic and scientific approach to investigating spoken input and performance with future language technology.

Military and government applications of human-machine communication by voice.

This paper describes a range of opportunities for military and government applications of human-machine communication by voice, based on visits and contacts with numerous user organizations in the United States. The applications include some that appear to be feasible by careful integration of current state-of-the-art technology and others that will require a varying mix of advances in speech technology and in integration of the technology into applications environments. Applications that are described include (1) speech recognition and synthesis for mobile command and control; (2) speech processing for a portable multifunction soldier's computer; (3) speech- and language-based technology for naval combat team tactical training; (4) speech technology for command and control on a carrier flight deck; (5) control of auxiliary systems, and alert and warning generation, in fighter aircraft and helicopters; and (6) voice check-in, report entry, and communication for law enforcement agents or special forces. A phased approach for transfer of the technology into applications is advocated, where integration of applications systems is pursued in parallel with advanced research to meet future needs.

Deployment of human-machine dialogue systems.

The deployment of systems for human-to-machine communication by voice requires overcoming a variety of obstacles that affect the speech-processing technologies. Problems encountered in the field might include variation in speaking style, acoustic noise, ambiguity of language, or confusion on the part of the speaker. The diversity of these practical problems encountered in the "real world" leads to the perceived gap between laboratory and "real-world" performance. To answer the question "What applications can speech technology support today?" the concept of the "degree of difficulty" of an application is introduced. The degree of difficulty depends not only on the demands placed on the speech recognition and speech synthesis technologies but also on the expectations of the user of the system. Experience has shown that deployment of effective speech communication systems requires an iterative process. This paper discusses general deployment principles, which are illustrated by several examples of human-machine communication systems.

Research in speech communication.

Advances in digital speech processing are now supporting application and deployment of a variety of speech technologies for human/machine communication. In fact, new businesses are rapidly forming about these technologies. But these capabilities are of little use unless society can afford them. Happily, explosive advances in microelectronics over the past two decades have assured affordable access to this sophistication as well as to the underlying computing technology. The research challenges in speech processing remain in the traditionally identified areas of recognition, synthesis, and coding. These three areas have typically been addressed individually, often with significant isolation among the efforts. But they are all facets of the same fundamental issue--how to represent and quantify the information in the speech signal. This implies deeper understanding of the physics of speech production, the constraints that the conventions of language impose, and the mechanism for information processing in the auditory system. In ongoing research, therefore, we seek more accurate models of speech generation, better computational formulations of language, and realistic perceptual guides for speech processing--along with ways to coalesce the fundamental issues of recognition, synthesis, and coding. Successful solution will yield the long-sought dictation machine, high-quality synthesis from text, and the ultimate in low bit-rate transmission of speech. It will also open the door to language-translating telephony, where the synthetic foreign translation can be in the voice of the originating talker.

A perspective on early commercial applications of voice-processing technology for telecommunications and aids for the handicapped.

The Colloquium on Human-Machine Communication by Voice highlighted the global technical community's focus on the problems and promise of voice-processing technology, particularly, speech recognition and speech synthesis. Clearly, there are many areas in both the research and development of these technologies that can be advanced significantly. However, it is also true that there are many applications of these technologies that are capable of commercialization now. Early successful commercialization of new technology is vital to ensure continuing interest in its development. This paper addresses efforts to commercialize speech technologies in two markets: telecommunications and aids for the handicapped.

Speech technology in the year 2001.

This paper introduces the session "Technology in the Year 2001" and is the first of four papers dealing with the future of human-machine communication by voice. In looking to the future it is important to recognize both the difficulties of technological forecasting and the frailties of the technology as it exists today--frailties that are manifestations of our limited scientific understanding of human cognition. The technology to realize truly advanced applications does not yet exist and cannot be supported by our presently incomplete science of speech. To achieve this long-term goal, the authors advocate a fundamental research program using a cybernetic approach substantially different from more conventional synthetic approaches. In a cybernetic approach, feedback control systems will allow a machine to adapt to a linguistically rich environment using reinforcement learning.

Toward the ultimate synthesis/recognition system.

This paper predicts speech synthesis, speech recognition, and speaker recognition technology for the year 2001, and it describes the most important research problems to be solved in order to arrive at these ultimate synthesis and recognition systems. The problems for speech synthesis include natural and intelligible voice production, prosody control based on meaning, capability of controlling synthesized voice quality and choosing individual speaking style, multilingual and multidialectal synthesis, choice of application-oriented speaking styles, capability of adding emotion, and synthesis from concepts. The problems for speech recognition include robust recognition against speech variations, adaptation/normalization to variations due to environmental conditions and speakers, automatic knowledge acquisition for acoustic and linguistic modeling, spontaneous speech recognition, naturalness and ease of human-machine interaction, and recognition of emotion. The problems for speaker recognition are similar to those for speech recognition. The research topics related to all these techniques include the use of articulatory and perceptual constraints and evaluation methods for measuring the quality of technology and systems.

The future of voice-processing technology in the world of computers and communications.

This talk, which was the keynote address of the NAS Colloquium on Human-Machine Communication by Voice, discusses the past, present, and future of human-machine communications, especially speech recognition and speech synthesis. Progress in these technologies is reviewed in the context of the general progress in computer and communications technologies.

An antagonistic vascular endothelial growth factor (VEGF) variant inhibits VEGF-stimulated receptor autophosphorylation and proliferation of human endothelial cells

Vascular endothelial growth factor (VEGF) is a potent mitogen with a unique specificity for endothelial cells and a key mediator of aberrant endothelial cell proliferation and vascular permeability in a variety of human pathological situations, such as tumor angiogenesis, diabetic retinopathy, rheumatoid arthritis, or psoriasis. VEGF is a symmetric homodimeric molecule with two receptor binding interfaces lying on each pole of the molecule. Herein we report on the construction and recombinant expression of an asymmetric heterodimeric VEGF variant with an intact receptor binding interface at one pole and a mutant receptor binding interface at the second pole of the dimer. This VEGF variant binds to VEGF receptors but fails to induce receptor activation. In competition experiments, the heterodimeric VEGF variant antagonizes VEGF-stimulated receptor autophosphorylation and proliferation of endothelial cells. A 15-fold excess of the heterodimer was sufficient to inhibit VEGF-stimulated endothelial cell proliferation by 50%, and a 100-fold excess resulted in an almost complete inhibition. By using a rational approach that is based on the structure of VEGF, we have shown the feasibility to construct a VEGF variant that acts as an VEGF antagonist.

The crystal structure of human interferon β at 2.2-Å resolution

Type I interferons (IFNs) are helical cytokines that have diverse biological activities despite the fact that they appear to interact with the same receptor system. To achieve a better understanding of the structural basis for the different activities of α and β IFNs, we have determined the crystal structure of glycosylated human IFN-β at 2.2-Å resolution by molecular replacement. The molecule adopts a fold similar to that of the previously determined structures of murine IFN-β and human IFN-α2b but displays several distinct structural features. Like human IFN-α2b, human IFN-β contains a zinc-binding site at the interface of the two molecules in the asymmetric unit, raising the question of functional relevance for IFN-β dimers. However, unlike the human IFN-α2b dimer, in which homologous surfaces form the interface, human IFN-β dimerizes with contact surfaces from opposite sides of the molecule. The relevance of the structure to the effects of point mutations in IFN-β at specific exposed residues is discussed. A potential role of ligand–ligand interactions in the conformational assembly of IFN receptor components is discussed.

Novel dimeric interface and electrostatic recognition in bacterial Cu,Zn superoxide dismutase

Eukaryotic Cu,Zn superoxide dismutases (CuZnSODs) are antioxidant enzymes remarkable for their unusually stable β-barrel fold and dimer assembly, diffusion-limited catalysis, and electrostatic guidance of their free radical substrate. Point mutations of CuZnSOD cause the fatal human neurodegenerative disease amyotrophic lateral sclerosis. We determined and analyzed the first crystallographic structure (to our knowledge) for CuZnSOD from a prokaryote, Photobacterium leiognathi, a luminescent symbiont of Leiognathid fish. This structure, exemplifying prokaryotic CuZnSODs, shares the active-site ligand geometry and the topology of the Greek key β-barrel common to the eukaryotic CuZnSODs. However, the β-barrel elements recruited to form the dimer interface, the strategy used to forge the channel for electrostatic recognition of superoxide radical, and the connectivity of the intrasubunit disulfide bond in P. leiognathi CuZnSOD are discrete and strikingly dissimilar from those highly conserved in eukaryotic CuZnSODs. This new CuZnSOD structure broadens our understanding of structural features necessary and sufficient for CuZnSOD activity, highlights a hitherto unrecognized adaptability of the Greek key β-barrel building block in evolution, and reveals that prokaryotic and eukaryotic enzymes diverged from one primordial CuZnSOD and then converged to distinct dimeric enzymes with electrostatic substrate guidance.

Free energy of burying hydrophobic residues in the interface between protein subunits

We have obtained an experimental estimate of the free energy change associated with variations at the interface between protein subunits, a subject that has raised considerable interest since the concept of accessible surface area was introduced by Lee and Richards [Lee, B. & Richards, F. M. (1971) J. Mol. Biol. 55, 379–400]. We determined by analytical ultracentrifugation the dimer–tetramer equilibrium constant of five single and three double mutants of human Hb. One mutation is at the stationary α1β1 interface, and all of the others are at the sliding α1β2 interface where cleavage of the tetramer into dimers and ligand-linked allosteric changes are known to occur. A surprisingly good linear correlation between the change in the free energy of association of the mutants and the change in buried hydrophobic surface area was obtained, after corrections for the energetic cost of losing steric complementarity at the αβ dimer interface. The slope yields an interface stabilization free energy of −15 ± 1.2 cal/mol upon burial of 1 Å2 of hydrophobic surface, in very good agreement with the theoretical estimate given by Eisenberg and McLachlan [Eisenberg, D. & McLachlan, A. D. (1986) Nature (London) 319, 199–203].

Engineering actin-resistant human DNase I for treatment of cystic fibrosis.

Human deoxyribonuclease I (DNase I), an enzyme recently approved for treatment of cystic fibrosis (CF), has been engineered to create two classes of mutants: actin-resistant variants, which still catalyze DNA hydrolysis but are no longer inhibited by globular actin (G-actin) and active site variants, which no longer catalyze DNA hydrolysis but still bind G-actin. Actin-resistant variants with the least affinity for actin, as measured by an actin binding ELISA and actin inhibition of [33P] DNA hydrolysis, resulted from the introduction of charged, aliphatic, or aromatic residues at Ala-114 or charged residues on the central hydrophobic actin binding interface at Tyr-65 or Val-67. In CF sputum, the actin-resistant variants D53R, Y65A, Y65R, or V67K were 10-to 50-fold more potent than wild type in reducing viscoelasticity as determined in sputum compaction assays. The reduced viscoelasticity correlated with reduced DNA length as measured by pulsed-field gel electrophoresis. In contrast, the active site variants H252A or H134A had no effect on altering either viscoelasticity or DNA length in CF sputum. The data from both the active site and actin-resistant variants demonstrate that the reduction of viscoelasticity by DNase I results from DNA hydrolysis and not from depolymerization of filamentous actin (F-actin). The increased potency of the actin-resistant variants indicates that G-actin is a significant inhibitor of DNase I in CF sputum. These results further suggest that actin-resistant DNase I variants may have improved efficacy in CF patients.

Crystal structure at 2.6-A resolution of human macrophage migration inhibitory factor.

Macrophage migration inhibitory factor (MIF) was the first cytokine to be described, but for 30 years its role in the immune response remained enigmatic. In recent studies, MIF has been found to be a novel pituitary hormone and the first protein identified to be released from immune cells on glucocorticoid stimulation. Once secreted, MIF counterregulates the immunosuppressive effects of steroids and thus acts as a critical component of the immune system to control both local and systemic immune responses. We report herein the x-ray crystal structure of human MIF to 2.6 angstrom resolution. The protein is a trimer of identical subunits. Each monomer contains two antiparallel alpha-helices that pack against a four-stranded beta-sheet. The monomer has an additional two beta-strands that interact with the beta-sheets of adjacent subunits to form the interface between monomers. The three beta-sheets are arranged to form a barrel containing a solvent-accessible channel that runs through the center of the protein along a molecular 3-fold axis. Electrostatic potential maps reveal that the channel has a positive potential, suggesting that it binds negatively charged molecules. The elucidated structure for MIF is unique among cytokines or hormonal mediators, and suggests that this counterregulator of glucocorticoid action participates in novel ligand-receptor interactions.