---
id: "phoneme"
slug: "phoneme"
name: "Phoneme"
domain: "Speech Processing & Acoustic Engineering"
taskType: "binary_classification"
numRows: 5404
numColumns: 6
targetColumn: "phoneme_sound_class"
primaryMetric: "1 - AUC Error"
primaryScore: "0.0278"
parquetUrl: "https://huggingface.co/datasets/carlahq/demo-tabular-benchmarks/resolve/main/data/phoneme/phoneme.parquet"
spreadsheetUrl: "https://docs.google.com/spreadsheets/d/1O95sBoK9VKvHX_CLEgAAZwfoDzafhPWBUvJAONzRq9M/edit?usp=sharing"
canonicalUrl: "https://carlahq.eu/datasets/phoneme"
---

# Dataset: Phoneme (phoneme)

> Benchmark dataset comprising 5,404 acoustic observations across 5 spectral harmonic features designed to classify speech vowel phonemes into nasal versus oral sound categories.

## Dataset Overview

| Attribute | Value |
| :--- | :--- |
| **Task Type** | `binary_classification` |
| **Business Domain** | Speech Processing & Acoustic Engineering |
| **Target Column** | `phoneme_sound_class` |
| **Rows (Instances)** | 5,404 |
| **Features (Columns)** | 6 |
| **Primary Evaluation Metric** | 1 - AUC Error |
| **Best Benchmark Score** | **0.0278** |

## Business Decision Context
**Context:** Analytical speech recognition pipelines and automated voice assistants require robust, speaker-independent phonetic feature extraction to accurately transcribe spoken language in real time under noisy conditions.
**Objective:** Classify acoustic vowel segments as either nasal or oral based on the energy-normalized amplitudes of their first five spectral harmonics.
**Decision Impact:** Accurate phoneme disambiguation prevents cascading phonetic transcription errors in automated speech-to-text and IVR systems, directly reducing downstream semantic misunderstanding and customer call escalation costs.

## Feature Schema
| Column Name | Data Type | Role | Distinct Values | Missing Values (%) | Description |
| :--- | :--- | :--- | :--- | :--- | :--- |
| `phoneme_sound_class` | `numeric` | target | 2 | 0 | Target phonetic articulation category representing nasal vowel sounds (Class 1) versus oral vowel sounds (Class 2). |
| `harmonic_1_normalized_amplitude` | `numeric` | feature | 5,336 | 0 | Normalized amplitude of the 1st spectral harmonic (AH1/Ene), signed positive for local spectral maxima and negative otherwise. |
| `harmonic_2_normalized_amplitude` | `numeric` | feature | 5,312 | 0 | Normalized amplitude of the 2nd spectral harmonic (AH2/Ene), signed positive for local spectral maxima and negative otherwise. |
| `harmonic_3_normalized_amplitude` | `numeric` | feature | 5,308 | 0 | Normalized amplitude of the 3rd spectral harmonic (AH3/Ene), signed positive for local spectral maxima and negative otherwise. |
| `harmonic_4_normalized_amplitude` | `numeric` | feature | 5,336 | 0 | Normalized amplitude of the 4th spectral harmonic (AH4/Ene), signed positive for local spectral maxima and negative otherwise. |
| `harmonic_5_normalized_amplitude` | `numeric` | feature | 4,499 | 0 | Normalized amplitude of the 5th spectral harmonic (AH5/Ene), signed positive for local spectral maxima and negative otherwise. |

## Empirical Benchmark Leaderboard
| Rank | Algorithm / Model | Model Family | Runtime | Accuracy | ROC AUC | F1-Score |
| :--- | :--- | :--- | :--- | :--- | :--- | :--- |
| 1 | **Google TabFM v1.0** | Tabular Foundation Model | PyTorch/Python | **92.98%** | 0.9798 | 0.8793 |
| 2 | **EXAONE Tabular** | Tabular Foundation Model | PyTorch/Python | **91.96%** | 0.9724 | 0.8617 |
| 3 | **TabPFN v3** | Tabular Foundation Model | PyTorch/Python | **91.95%** | 0.9726 | 0.8609 |
| 4 | **Carla Engine (TabICLv2 WebGPU)** | Tabular Foundation Model | Browser | **91.78%** | 0.9721 | 0.8586 |
| 5 | **Streaming nanotabicl** | Tabular Foundation Model | PyTorch/Python | **91.77%** | 0.9722 | 0.8578 |
| 6 | **nanotabicl Vanilla** | Tabular Foundation Model | PyTorch/Python | **91.77%** | 0.9722 | 0.8583 |
| 7 | **KStar** | Machine Learning Model | Python | **91.73%** | 0.9674 | 0.9162 |
| 8 | **Full TabICLv2 (PyTorch Reference)** | Tabular Foundation Model | PyTorch/Python | **91.72%** | 0.9722 | 0.8573 |
| 9 | **RandomForest** | Random Forest | Python | **91.52%** | 0.9643 | 0.9149 |
| 10 | **AttributeSelectedClassifier RandomForest** | Random Forest | Python | **91.21%** | 0.9622 | 0.9118 |
| 11 | **Bagging RandomForest** | Random Forest | Python | **90.88%** | 0.9623 | 0.9085 |
| 12 | **RandomCommittee RandomTree** | Decision Tree | Python | **90.71%** | 0.9512 | 0.9062 |
| 13 | **IB1** | Machine Learning Model | Python | **90.43%** | 0.8761 | 0.9035 |
| 14 | **AdaBoostM1 IBk** | AdaBoost | Python | **90.43%** | 0.8763 | 0.9035 |
| 15 | **Bagging IBk** | k-Nearest Neighbors (k-NN) | Python | **90.10%** | 0.9281 | 0.9000 |

### Repeated Cross-Validation Fold Scores (TabICLv2 Reference)
| Fold | Train Samples | Test Samples | Metric Score | Latency |
| :--- | :--- | :--- | :--- | :--- |
| Fold 0 | 3602 | 1802 | 0.0304 | 1375ms |
| Fold 1 | 3603 | 1801 | 0.0291 | 1360ms |
| Fold 2 | 3603 | 1801 | 0.0231 | 1369ms |
| Fold 0 | 3602 | 1802 | 0.0268 | 1352ms |
| Fold 1 | 3603 | 1801 | 0.0233 | 1373ms |
| Fold 2 | 3603 | 1801 | 0.0269 | 1363ms |
| Fold 0 | 3602 | 1802 | 0.0270 | 1365ms |
| Fold 1 | 3603 | 1801 | 0.0341 | 1367ms |
| Fold 2 | 3603 | 1801 | 0.0300 | 1365ms |

## Provenance & Downloadable Artifacts
- **Dataset Parquet File:** [Direct Parquet Download](https://huggingface.co/datasets/carlahq/demo-tabular-benchmarks/resolve/main/data/phoneme/phoneme.parquet)
- **Google Sheets Interactive Demo:** [Open in Google Sheets](https://docs.google.com/spreadsheets/d/1O95sBoK9VKvHX_CLEgAAZwfoDzafhPWBUvJAONzRq9M/edit?usp=sharing)
- **OpenML Source:** [OpenML Dataset](https://www.openml.org/d/1489)
- **Academic Paper / Citation:** [Read Research Paper](https://openml.org/d/1489)
- **Canonical HTML Page:** [https://carlahq.eu/datasets/phoneme](https://carlahq.eu/datasets/phoneme)