Speech and audio models are AI models trained to understand, transform, or generate spoken language and sound.
Speech and audio models handle different sound tasks.
These are related capabilities, but they are not one interchangeable process.
A group of specialists working with sound:
A meeting recording becomes searchable text.
A system recognizes speech and responds with spoken audio.
Written content becomes a natural-sounding voice.
A prompt becomes a new musical phrase or sound effect.
Speech models: A broad family for speech tasks.
TTS: One task that turns text into speech.
Audio model: Works with sound representations or waveforms.
LLM: Primarily works with language tokens.
Recognition: Usually transcribes spoken words.
Understanding: May also require language, context, intent, and emotion analysis.
Multimodal Models can combine audio with text and images.
Explore next
Speech and audio models turn sound into useful representations, text, or newly generated audio depending on the task.
A video will be added when a real explainer asset is available.