--- license: apache-2.0 language: - ky - ru task_categories: - tabular-classification - audio-classification tags: - keyword-spotting - wake-word - kyrgyz - speech - mfcc - spectral-features - educational - teaching size_categories: - 10K **One-line summary.** 40 000 clips → 250 acoustic features → binary label > (`1` = wake word, `0` = not). Mildly imbalanced (1 : 3). Built for an ML course that has > just covered linear models (Logistic Regression, SVM) and is about to meet trees. --- ## Table of contents 1. [The task: what is Keyword Spotting?](#1-the-task-what-is-keyword-spotting) 2. [A short history of wake words](#2-a-short-history-of-wake-words) 3. [Where the audio comes from (provenance)](#3-where-the-audio-comes-from-provenance) 4. [From a waveform to a feature vector](#4-from-a-waveform-to-a-feature-vector) 5. [The features, in detail (with formulas)](#5-the-features-in-detail-with-formulas) 6. [Temporal aggregation: variable-length audio → fixed vector](#6-temporal-aggregation-variable-length-audio--fixed-vector) 7. [Dataset schema](#7-dataset-schema) 8. [Modeling challenges (read this before you trust your accuracy)](#8-modeling-challenges) 9. [License, provenance & citation](#9-license-provenance--citation) --- ## 1. The task: what is Keyword Spotting? **Keyword Spotting (KWS)** is the problem of detecting a small set of predefined words or short phrases in an audio stream. The most familiar special case is **wake-word detection** (also *hotword* or *trigger-word* detection): a tiny, always-listening model that waits for a single phrase — *"Hey Siri"*, *"OK Google"*, *"Alexa"* — and only then wakes up the heavy, cloud-based speech recogniser. Formally, given an audio segment $x(t)$ we want a decision function $$ \hat{y} = \mathbb{1}\!\left[\, p(\text{keyword} \mid x) \ge \tau \,\right], $$ where $\tau$ is an operating threshold. In this dataset the keyword is the Kyrgyz given name **«Акылай»** (three syllables, stress on the final *-ай*), and the task is reduced to its cleanest form: **binary classification of pre-segmented 2-second-scale clips** — keyword vs. non-keyword. KWS has several properties that make it a richer teaching example than tabular toy datasets: - **Strong class imbalance.** In deployment the keyword is vanishingly rare (a wake word may fire a handful of times per day against hours of non-keyword audio). Here we use a gentle **1 : 3** ratio — enough to make *accuracy misleading* without being degenerate. - **Asymmetric error costs.** A *false reject* (missing the keyword) annoys the user once; a *false accept* (waking up on a TV advert) is far worse. This motivates the whole precision/recall/threshold toolkit rather than a single accuracy number. - **A meaningful feature-engineering step.** Audio is not naturally tabular. Turning a waveform into a fixed-length vector is itself a modelling decision — and a great lesson. --- ## 2. A short history of wake words - **1950s–60s — first isolated-word recognisers.** Bell Labs' *Audrey* (1952) recognised spoken digits from a single speaker; IBM's *Shoebox* (1962) handled 16 words. These were analog/template machines, but they established the core idea of matching short acoustic patterns. - **1970s–80s — features and dynamic time warping.** The **cepstrum** and then **Mel-Frequency Cepstral Coefficients (MFCCs)** (Davis & Mermelstein, 1980) became the standard front-end, and **DTW** allowed matching words of different durations. - **1980s–2000s — statistical models.** **Hidden Markov Models (HMMs)** with Gaussian Mixture emissions dominated speech. Classic KWS was often **keyword-filler HMMs**: one model for the keyword, a "garbage" model for everything else. - **2014 — the deep-learning turning point for KWS.** Google's *"Small-footprint keyword spotting using deep neural networks"* (Chen, Parada & Heigold, 2014) showed a compact DNN on log-mel features beating the HMM pipeline — the recipe behind *"OK Google"* on-device. - **2014–2017 — the smart-speaker era.** *Amazon Echo / Alexa* (2014), *"Hey Siri"* on a dedicated low-power core (2017), and *"OK Google"* turned always-on wake-word detection into a mass-market component. Constraints became extreme: a few **tens of kilobytes** of parameters, running continuously at milliwatts. - **2018–present — convolutional & streaming models.** Architectures such as **TC-ResNet**, **BC-ResNet**, and depthwise-separable CNNs pushed accuracy up while keeping the model tiny enough for an MCU/NPU. This dataset's parent project trains exactly such a tiny on-device model (a ~30 K-parameter BC-ResNet) for «Акылай». **The features you have here are the *classical* front-end** — MFCCs and spectral descriptors — which is both historically faithful and a perfect bridge from "linear models on tables" to "real speech". --- ## 3. Where the audio comes from (provenance) The clips come from **four different sources**, recorded in the `source` column. The single most important thing to understand about this dataset is the split between **synthetic (text-to-speech, TTS)** audio and **real human** audio. | `source` | `label` | n | Synthetic / real | What it is | |---|:---:|---:|---|---| | `positive` | **1** | 10 000 | **TTS** (in-house Kyrgyz TTS) | The wake word **«Акылай»**, spoken by a Kyrgyz text-to-speech model trained on podcast voices. | | `base_neg` | 0 | 7 680 | **TTS** (*same* engine as positives) | Other Kyrgyz words/phrases from the **same** TTS voice — *not* the wake word. | | `confusable` | 0 | 2 949 | **TTS** (KaniTTS, a *different* engine) | Phonetic near-neighbours — words ending in *-ай / -лай / -кай / -бай* (e.g. *Алтынай, калай, чай, лайк*) designed to look "almost" like the keyword. | | `podcast` | 0 | 19 371 | **Real human speech** | 2-second cuts of continuous Kyrgyz podcast speech — natural, spontaneous, with no wake word. | **Why this matters (and why we keep `source`).** The negatives are not homogeneous: - `podcast` is **real, out-of-domain** audio → in practice it is *easy* to separate from the synthetic positives, partly for the wrong reason (the model can latch onto "synthetic vs. real" timbre rather than the word itself — a classic **shortcut**). - `base_neg` shares the **exact same TTS voice** as the positives, so the *only* thing distinguishing it from a positive is the **word** → this is the **honest, hard** part of the problem. - `confusable` tests robustness to **phonetically similar words**. The `source` column is **not a feature** — never feed it to the model. It is provided for **error analysis**: *which kind of negative does your model actually fail on?* (Spoiler from our baseline experiments: almost all false positives come from `base_neg`.) > **Wake word.** «Акылай» — a Kyrgyz feminine given name, 3 syllables, stress on `-ай`. > All audio is 16 kHz mono. --- ## 4. From a waveform to a feature vector A raw clip is a sequence of $16\,000$ amplitude samples per second — far too high-dimensional and variable-length to feed to a classifier directly. The standard speech front-end turns it into a compact, fixed-length descriptor in three stages: **framing**, **time–frequency transform**, and **feature extraction + aggregation**. ### 4.1 Framing The signal $x(n)$ is cut into overlapping short frames so that each frame is approximately **stationary** (the vocal tract changes slowly, ~10 ms scale). We use $$ N_{\text{fft}} = 512 \;(\approx 32\,\text{ms window}), \qquad H = 160 \;(\approx 10\,\text{ms hop}), $$ with a **Hann window** $w(n) = 0.5\left(1 - \cos\frac{2\pi n}{N-1}\right)$ to reduce spectral leakage. ### 4.2 Short-Time Fourier Transform (STFT) For frame index $m$ and frequency bin $k$, $$ X(m,k) \;=\; \sum_{n=0}^{N_{\text{fft}}-1} x(n + mH)\, w(n)\, e^{-\,j\,2\pi k n / N_{\text{fft}}}. $$ From it we form the **magnitude** $|X(m,k)|$ and **power** spectrogram $$ S(m,k) = |X(m,k)|^2 . $$ The bin $k$ corresponds to physical frequency $f_k = \dfrac{k}{N_{\text{fft}}}\, f_s$, with $f_s = 16\,000$ Hz. Everything below is computed **per frame** $m$ and then aggregated over time (§6). --- ## 5. The features, in detail (with formulas) We extract **12 groups** of descriptors. They fall into three families: - **Cepstral** (MFCC + dynamics) — compact model of the spectral *envelope* (the vocal-tract shape that defines phonemes). - **Spectral shape** (centroid, bandwidth, roll-off, flatness, contrast) — interpretable scalar summaries of *where* and *how* spectral energy is distributed ("timbre"). - **Energy / harmonicity / time-domain** (RMS, chroma, ZCR) — loudness, pitch-class content, and a rough voicing/noisiness cue. ### 5.1 Mel filterbank and log-mel energies Human pitch perception is roughly logarithmic, captured by the **mel scale** $$ \text{mel}(f) = 2595 \,\log_{10}\!\left(1 + \frac{f}{700}\right). $$ We place $M = 40$ overlapping **triangular filters** $H_j(k)$ equally spaced on the mel axis and integrate the power spectrum through them: $$ E_{\text{mel}}(m,j) \;=\; \sum_{k} H_j(k)\, S(m,k), \qquad j = 1,\dots,40 . $$ The dataset stores the **log (dB) mel energies** $\;10\log_{10} E_{\text{mel}}(m,j)\;$ (columns `mel0…mel39`). *Interpretation:* a coarse, perceptually-warped picture of the spectral envelope — high values in low-mel bands mean energy concentrated at low pitch, etc. ### 5.2 Mel-Frequency Cepstral Coefficients (MFCC) MFCCs decorrelate the log-mel vector with a **Discrete Cosine Transform (DCT-II)**: $$ c(m,i) \;=\; \sum_{j=1}^{M} \log E_{\text{mel}}(m,j)\, \cos\!\left[\frac{\pi i}{M}\left(j - \tfrac{1}{2}\right)\right], \qquad i = 0,\dots,19 . $$ We keep the first **20** coefficients (`mfcc0…mfcc19`). Low-order coefficients capture the **smooth spectral envelope** (formant structure → which phoneme is spoken); $c(m,0)$ is proportional to overall log-energy. MFCCs are the single most important speech feature historically and are nearly **uncorrelated**, which suits linear models well. ### 5.3 Delta ($\Delta$) and delta-delta ($\Delta\Delta$) coefficients A single frame says nothing about *motion*. The **delta** features approximate the temporal derivative of each coefficient with a regression over $\pm\Theta$ frames: $$ \Delta c(m,i) \;=\; \frac{\displaystyle\sum_{\theta=1}^{\Theta} \theta\,\bigl[c(m+\theta,i) - c(m-\theta,i)\bigr]} {\displaystyle 2\sum_{\theta=1}^{\Theta} \theta^{2}} . $$ The **delta-delta** (acceleration) features are the deltas of the deltas. Columns `d1_0…d1_19` ($\Delta$) and `d2_0…d2_19` ($\Delta\Delta$). *Interpretation:* how fast the spectrum is changing — crucial for distinguishing a short, dynamic spoken word from quasi-stationary noise or music. ### 5.4 Spectral centroid The "centre of mass" of the spectrum — a strong correlate of perceived **brightness**: $$ \text{centroid}(m) \;=\; \frac{\sum_{k} f_k \,|X(m,k)|}{\sum_{k} |X(m,k)|}. $$ ### 5.5 Spectral bandwidth The spread of energy around the centroid (here the $p=2$, i.e. standard-deviation, form): $$ \text{bandwidth}(m) \;=\; \left( \frac{\sum_{k} |X(m,k)|\,\bigl(f_k - \text{centroid}(m)\bigr)^{2}} {\sum_{k} |X(m,k)|} \right)^{1/2}. $$ ### 5.6 Spectral roll-off The frequency $f_R(m)$ below which a fraction $\rho = 0.85$ of the total spectral energy lies: $$ f_R(m) = \min\Big\{ f_K \;:\; \sum_{k:\,f_k \le f_K} |X(m,k)| \;\ge\; \rho \sum_{k} |X(m,k)| \Big\}. $$ Separates voiced/low-frequency-dominated frames from broadband/fricative ones. ### 5.7 Spectral flatness The ratio of the **geometric** to the **arithmetic** mean of the power spectrum — a *tonality vs. noisiness* measure in $[0,1]$: $$ \text{flatness}(m) \;=\; \frac{\exp\!\left(\frac{1}{K}\sum_{k}\ln S(m,k)\right)}{\frac{1}{K}\sum_{k} S(m,k)}. $$ A value near $1$ ⇒ white-noise-like (flat); near $0$ ⇒ tonal/harmonic (peaky), as in voiced speech. Very useful for telling speech from noise/music. ### 5.8 Spectral contrast For each of **7 sub-bands** $b$, the (log) difference between the strongest **peaks** and the weakest **valleys** in that band: $$ \text{contrast}(m,b) \;=\; \overline{\text{Peak}}_b(m) \;-\; \overline{\text{Valley}}_b(m), $$ where the peak/valley terms are the mean log-energies of the top/bottom quantile of bins in band $b$. High contrast ⇒ clear harmonic structure (formants stand out over the noise floor). Columns `contrast0…contrast6`. *In our baseline this was the single most informative group.* ### 5.9 Chroma Projects the spectrum onto the **12 pitch classes** of the equal-tempered scale: $$ \text{chroma}(m,p) \;=\; \sum_{k\,:\,\text{pitch-class}(f_k)=p} |X(m,k)|, \qquad p = 0,\dots,11 . $$ Octave-invariant harmonic content. Borrowed from music IR; for speech it is a weak but non-trivial cue. Columns `chroma0…chroma11`. ### 5.10 Zero-Crossing Rate (ZCR) A cheap time-domain measure of how often the waveform changes sign within a frame of length $L$: $$ \text{ZCR}(m) \;=\; \frac{1}{2L}\sum_{n}\bigl|\operatorname{sgn} x(n) - \operatorname{sgn} x(n-1)\bigr|. $$ High ZCR ⇒ noisy/fricative/unvoiced; low ZCR ⇒ voiced/low-frequency. A classic voicing proxy. ### 5.11 Root-Mean-Square energy (RMS) Per-frame loudness: $$ \text{RMS}(m) \;=\; \sqrt{\frac{1}{L}\sum_{n} x(n)^{2}} . $$ Tracks the speech envelope (syllable onsets/offsets, silences). --- ## 6. Temporal aggregation: variable-length audio → fixed vector The features above produce **one value per frame**, so a clip is a *matrix* $\phi(m)$ of shape (features × frames), and clips have **different numbers of frames** (different durations). Classical models need a **fixed-length** vector. We therefore summarise each per-frame feature $\phi$ over its $T$ frames by its **mean** and **standard deviation**: $$ \mu_\phi = \frac{1}{T}\sum_{m=1}^{T} \phi(m), \qquad \sigma_\phi = \sqrt{\frac{1}{T}\sum_{m=1}^{T}\bigl(\phi(m) - \mu_\phi\bigr)^{2}} . $$ Hence every group contributes **two** columns per channel (`…_mean`, `…_std`), and *every clip maps to the same 250-dimensional vector* regardless of length. The `std` summaries turn out to be very informative — they encode *how much the spectrum moves over time*, which separates a short spoken word from stationary background. (In our baseline, several `…_std` features rank at the very top of tree-based feature importances.) > **Design choice / leakage note.** Clip **duration is deliberately *not* a feature.** In the > raw corpus all `podcast` negatives are exactly 2.000 s while positives are shorter, so > duration alone would let a model "cheat". Aggregating to mean/std makes the descriptor > length-invariant and removes that shortcut. (The deeper *synthetic-vs-real* shortcut remains > — see §8.) ### Feature budget (250 total) | Group | Per-frame channels | × {mean, std} | Columns | |---|---:|---:|---:| | MFCC | 20 | 2 | 40 | | $\Delta$ MFCC | 20 | 2 | 40 | | $\Delta\Delta$ MFCC | 20 | 2 | 40 | | log-mel energies | 40 | 2 | 80 | | spectral contrast | 7 | 2 | 14 | | chroma | 12 | 2 | 24 | | centroid | 1 | 2 | 2 | | bandwidth | 1 | 2 | 2 | | roll-off | 1 | 2 | 2 | | flatness | 1 | 2 | 2 | | ZCR | 1 | 2 | 2 | | RMS | 1 | 2 | 2 | | **Total** | | | **250** | --- ## 7. Dataset schema One parquet file, **40 000 rows × 252 columns**, all features `float32`, no missing values. | Column(s) | Type | Role | Notes | |---|---|---|---| | `mfcc{0..19}_{mean,std}` | float32 | feature | cepstral envelope | | `d1_{0..19}_{mean,std}` | float32 | feature | $\Delta$ MFCC | | `d2_{0..19}_{mean,std}` | float32 | feature | $\Delta\Delta$ MFCC | | `mel{0..39}_{mean,std}` | float32 | feature | log-mel energies (dB) | | `contrast{0..6}_{mean,std}` | float32 | feature | spectral contrast | | `chroma{0..11}_{mean,std}` | float32 | feature | pitch-class energy | | `centroid_{mean,std}` | float32 | feature | brightness | | `bandwidth_{mean,std}` | float32 | feature | spectral spread | | `rolloff_{mean,std}` | float32 | feature | 85 % roll-off freq. | | `flatness_{mean,std}` | float32 | feature | tonality | | `zcr_{mean,std}` | float32 | feature | zero-crossing rate | | `rms_{mean,std}` | float32 | feature | loudness | | **`label`** | int | **target** | `1` = «Акылай», `0` = negative | | **`source`** | string | **metadata** | `positive` / `base_neg` / `confusable` / `podcast` — **do not train on this** | **Class balance:** 10 000 positive / 30 000 negative (**25 % positive, 1 : 3**). --- ## 8. Modeling challenges This is where the dataset earns its keep as a teaching tool. Things students *should* run into: 1. **Class imbalance → accuracy lies.** A constant "always negative" classifier already scores **75 % accuracy** but is useless (F1 = 0, PR-AUC = 0.25). Insist on **precision, recall, F1, ROC-AUC, and especially PR-AUC**, plus the confusion matrix. Tools to discuss: `class_weight="balanced"`, threshold tuning, resampling. 2. **The synthetic-vs-real shortcut (the big one).** Positives are TTS; `podcast` negatives are real human speech. A model can score deceptively well by learning *"synthetic timbre vs. real"* instead of *"the word Akylai vs. other words"*. The antidote is **per-`source` error analysis**: evaluate false-positive rate **separately** on `podcast`, `base_neg`, and `confusable`. The honest difficulty lives in `base_neg` (same TTS voice, different word). 3. **Operating point & asymmetric costs.** F1 is not the deployment metric; real KWS cares about *recall at a fixed false-alarm rate*. Sweep the threshold $\tau$ and read off the precision/recall trade-off — a natural lead-in to ROC and PR curves. 4. **Feature scaling matters — but only for some models.** Distance/margin-based learners (Logistic Regression, SVM, k-NN) need `StandardScaler`; tree ensembles (Random Forest, Gradient Boosting) are scale-invariant. A clean side-by-side lesson. 5. **High dimensionality & redundancy.** 250 features, many correlated (MFCC vs. mel; mean vs. std). Good ground for regularisation ($L_1/L_2$), feature importance, and dimensionality reduction. Note that in a 2-D **PCA** projection the classes **overlap heavily** — a useful reminder that "I can't see a boundary in 2-D" does *not* mean the classes are inseparable in 250-D. 6. **Speaker / generator leakage.** Positives come from a limited set of synthetic voices; a purely random train/test split can leak speaker identity and **inflate** scores. A stricter evaluation would split by speaker or by source. Worth at least *discussing*. 7. **It's "easy enough" to be encouraging, hard enough to be real.** Even a linear model reaches high PR-AUC on these features, so beginners get a rewarding result quickly — while the per-source breakdown leaves a genuine, interpretable hard core to dig into. --- ## 9. License, provenance & citation - **License:** Apache-2.0. - **Languages:** Kyrgyz (`ky`), with some Russian (`ru`) confusables. - **Audio provenance:** positives and `base_neg` are generated by an in-house Kyrgyz text-to-speech model (trained on podcast voices); `confusable` clips are generated by **KaniTTS**; `podcast` negatives are 2-second cuts of real Kyrgyz-language podcast speech. Features were extracted with `librosa` (16 kHz, $N_{\text{fft}}=512$, hop $=160$). - **Parent project:** the «Акылай» on-device wake-word detector (`KaniTTS-research-team/AkylAi_Wake_Word_V4`). This features table is a derived, tabular **teaching** snapshot — it does **not** contain audio. ```bibtex @misc{akylai_kws_features, title = {Akylai KWS Features: An Educational Spectral-Feature Dataset for Keyword Spotting}, author = {AkylAi Wake Word project}, year = {2026}, note = {Derived tabular features (MFCC + spectral descriptors) over the Akylai wake-word corpus}, license = {Apache-2.0} } ``` > **Educational use.** This dataset is intended for teaching classification and audio feature > engineering. The synthetic positives and the domain split between TTS and real speech make it > unsuitable as-is for benchmarking a production wake-word detector.