Back to corpus
research noteexperiment writeup candidatescore 18
CC-Speak
- **Lock-free audio capture** using CPAL - **Real-time RMS/peak metering** for UI feedback - **WAV encoding** with quality metrics (SNR, clipping detection) - **Python bindings** via PyO3
Full HTML reader
Read the full artifact
Extracted abstract or opening context
- **Lock-free audio capture** using CPAL - **Real-time RMS/peak metering** for UI feedback - **WAV encoding** with quality metrics (SNR, clipping detection) - **Python bindings** via PyO3
| Metric | Python (PyAudio) | Rust (CC-Speak) | |--------|------------------|-----------------| | Recording start latency | ~100ms | <10ms | | WAV encoding time | ~50ms | <5ms | | Total to clipboard | ~300ms | <50ms |
| Metric | Description | |--------|-------------| | `rms_energy` | Root mean square (average loudness) 0.0-1.0 | | `peak_amplitude` | Maximum sample value 0.0-1.0 | | `snr_db` | Estimated signal-to-noise ratio in dB | | `has_clipping` | Whether audio clipping was detected | | `quality_score` | Composite quality score 0.0-1.0 |
| Score | Level | Description | |-------|-------|-------------| | 0.9+ | Excellent | Studio quality, clear speech | | 0.7-0.9 | Good | Clear speech, minor noise | | 0.5-0.7 | Acceptable | Some noise, speech intelligible | | 0.3-0.5 | Poor | Significant noise | | <0.3 | Reject | Too noisy for training |
- **cpal** - Cross-platform audio I/O - **hound** - WAV encoding - **pyo3** - Python bindings (optional) - **cc-core-rs** - Core infrastructure
Promotion decision
What has to happen next
Attach run IDs, datasets, metrics, and reproduction commands.
Why this is not always a full paper yet
Corpus pages are public-safe readers for discovered workspace artifacts. They are not automatically final papers. A corpus item becomes a polished paper only after the editable source, evidence checkpoints, references, figures, render path, and release status are attached through the paper schema.