---
title: "Utility audio, speech, vectors, and text"
lesson_id: "05"
---

# Utility audio, speech, vectors, and text

- **Lesson ID:** 05
- **Goal:** Explain ASR, embedding, and encoding utilities by their input and output jobs.
- **Human lesson:** [05-hear-and-embed.html](05-hear-and-embed.html)

## Prerequisites

- Lesson 01 and the idea that utility is a set of focused jobs.
- The language-versus-image difference in [the architectures course](../model-architectures-plain/).

## Inputs, outputs, and artifacts

- **Inputs:** Speech, image, audio, or text and the desired representation or conversion.
- **Outputs:** A simple map of ASR, embeddings, and text encoding utilities.
- **Artifacts:** Progressive `.flow` diagrams for speech, vectors, and text.

## Agent build steps

1. Define ASR as speech to text.
2. Name `whisper-large-v3-turbo`, `whisper-large-v3`, and `wav2vec2-base` exactly.
3. Define embeddings as comparable vectors and name `clip-vit-b32` for image/text vectors and `clap-htsat` for audio vectors.
4. Define encoding and name `roberta-base`, `t5-small`, `t5-base`, and `t5-large`; describe T5 as sequence-to-sequence style.
5. Contrast these focused outputs with an LLM's broader language writing.
6. Link the architectures, forward, and From the metal to the model.

## Constraints

Use only the supplied utility names. Keep vector talk non-mathematical. No math, CUDA, quizzes, Mermaid, or em dashes.

## Key concepts

- ASR turns speech into text: `whisper-large-v3-turbo`, `whisper-large-v3`, `wav2vec2-base`.
- Embeddings return comparable vectors: `clip-vit-b32` for image and text, `clap-htsat` for audio.
- `roberta-base` encodes text. `t5-small`, `t5-base`, and `t5-large` change text sequence to sequence.
- These are focused jobs, not general chat like an `llm`.

## Takeaways

- You can pick ASR when the app needs words from speech.
- You can pick an embedding or encoding preset when the job is compare, represent, or change text.
- You can say when an `llm` fits better than ASR, embedding, or encoding.

## Acceptance checks

- All ASR, embedding, and encoding names are present exactly.
- Speech-to-text, image/text vectors, audio vectors, and text transformation are distinct.
- T5 is described as sequence-to-sequence style and RoBERTa as text encoding.
- Progressive `.flow` diagrams, prior-course links, and the Agent brief link are present.