Audio Training

Audio training covers speech-to-text, classification, and related audio-language tasks.

Dashboard

Open Train, choose Audio, then choose the audio method. Pick a task such as ASR before launching. Preflight reports missing dependencies or unsupported model families.

CLI

halo-forge audio train --dataset librispeech --model openai/whisper-small --task asr --output ~/.halo-forge/runs/audio-asr

Audio runs are dependency-sensitive; install the audio extras before scaling beyond smoke tests.