Boson AI co-founder Alex Smola talks with Sam Charrington about making voice interactionsA voice interface lets people interact with software through spoken input and audible responses. feel natural, from interruption handling to balancing audio token rates, model size, latency and costInference is the use of a trained model to produce an output from supplied input..
Smola describes adapting language models for audio, processing large audio datasets and combining conversational responses with background reasoning and tool calls. His company’s benchmark and cost claims are presented as his own.
They discuss measuring interaction qualityEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria., emotional context and personalizationAI personalization adapts information, recommendations, interfaces, or agent behavior to an individual or group using permitted context and observed needs., along with the additional challenges of keeping audiovisual avatars consistent in real time.
Watch on YouTube
