Sam Wake Word Audio Clips for Keyword Spotting

Contains audio clips for training a model to recognize the keyword 'Sam'. Each clip is labeled as positive (contains 'Sam') or negative (phonetically similar words). The dataset includes varied speaking styles, speeds, and intonations.

Use Cases

• Train an audio classification model to detect the 'Sam' wake word from labeled positive and negative clips.
• Analyze performance of keyword spotting models on phonetically similar negative examples.
• Benchmark wake-word detection systems using varied speaking styles and intonations.

Strengths

• Includes phonetically similar negative examples to improve model robustness.
• Positive clips feature varied speaking styles, speeds, and intonations.
• Designed for the specific task of wake-word detection for the keyword 'Sam'.

Limitations

• Unknown total number of audio clips or dataset size.
• Geographic and demographic coverage is unspecified, potentially limiting generalizability.
• Audio generation and collection methodology details are not fully provided.

Provenance

Source

quo-labs on Hugging Face.

Collection Method

Audio clips generated and labeled for the keyword 'Sam'; specific generation method not detailed.

Freshness

Last updated on 2026-03-24.

Geography

Region tag indicates 'us', but specific coverage is unknown.