Add audio text to text

[Dhiraj309]

This PR introduces a new audio-text-to-text task to the huggingface.js library. It enables converting audio input into text using automatic speech recognition (ASR) models.

Key updates:

Added packages/tasks/src/tasks/audio-text-to-text/ with:

data.ts – metadata for datasets, models, metrics, and demo.

inference.ts – logic for converting audio to text.

about.md – task description.

spec/input.json & spec/output.json – example inputs and expected outputs.

Task summary: "Convert audio input into text using speech-to-text (ASR) models."

Demonstration includes a sample .wav file and expected transcription output.

⚠️ Note: This task is currently not integrated into the main pipeline or automated tests. Manual testing is recommended before pipeline inclusion.