Fact · Speech recognition
In a hybrid architecture, a feedforward neural network's outputs are used as the probability estimates for an HMM-based architecture.
Checked against its source on October 17, 2026
In the source's own words
“whose outputs are used as probability estimates for an HMM-based architecture”
Speech and Language Processing, Chapter 16: Automatic Speech Recognition (draft), Daniel Jurafsky and James H. Martin, Stanford University, online manuscript of the third edition, 2026, Chapter 16, history of automatic speech recognition.
Open the original- What it measures
- The textbook lists the hybrid HMM/MLP architecture among the neural alternatives to HMM/GMM systems that appeared around 1990, and notes that the hybrid systems performed close to standard HMM/GMM models but were slow to train on the CPUs of that era.
- What it does not show
- A description of the architecture as the textbook defines it. Hybrid systems were a research design before they became standard; the dates given are approximate.