---

name: audio-data description: "Use SpeechRecognition 3.17.0 file audio loading and AudioData conversion/splitting without microphones or transcription services." disable-model-invocation: true metadata: disco-role: operating license: NOASSERTION

audio-data

Use this sub-skill for local file-based audio ingestion and speech_recognition.AudioData manipulation in SpeechRecognition 3.17.0.

Read these bundled references before acting:

• API reference for AudioFile, AudioData.from_file, segmenting, splitting, byte formats, conversion rules, and FLAC converter behavior.
• Workflows for copy-pasteable local-file conversion, chunking, and in-memory loading patterns.
• Troubleshooting for unsupported WAV/AIFF/FLAC files, converter failures, 24-bit/sample-width edge cases, split validation errors, and librosa/numpy SetupError handling.

Bundled helper scripts:

• scripts/audio_convert.py: convert a local input audio file into RAW/WAV/AIFF/FLAC outputs, optionally segmenting and splitting first.
• scripts/audio_smoke.py: generate tiny synthetic WAV data and verify AudioData, AudioData.from_file, AudioFile, conversions, and split() without source-repo fixtures.

Route away from this sub-skill when the task is not local audio-data handling:

• Any recognize_* transcription/API/model call belongs to the recognition-engines sub-skill.
• Microphone capture, ambient-noise calibration, listen(), and background listening belong to capture-listening.
• sprc CLI usage, model downloads, and model cache/setup issues belong to cli-model-setup.