A source model mitigation technique for distributed speech recognition over lossy packet channels | Publicación