AI Termcirca 1980· Added Jun 8, 2026
Acoustic Model
An acoustic model represents the relationship between audio signals and phonetic units.
Acoustic models are crucial in speech recognition systems, serving as a bridge between raw audio data and linguistic information. They use statistical representations of sounds to map audio signals to phonetic units (e.g., phonemes). These models are trained on large datasets of recorded speech to improve accuracy in recognizing spoken words under various conditions. Acoustic models are often paired with language models to enhance overall speech recognition performance.
Examples
- Google's speech recognition system uses acoustic models to transcribe spoken queries.
- Acoustic models help in improving the accuracy of call center transcription services.
- Used in mobile voice assistants to better understand user commands.
Common misconceptions
- Acoustic models alone can understand speech; they need language models for context.
- They are not responsible for speaker identification—that requires additional processing.
- Acoustic models are not static; they require constant updates with new data.
Related terms
Want more like this?
Open the full library
Fresh AI mastery content every 2 hours.