prokelly/neuromoyo-sahara-codeswitch-benchmark

NEUROMOYO — Sahara CodeSwitch Africa Benchmark

🔗 Live Benchmark Results

Interactive benchmark: https://www.neuromoyo.app/benchmark

This page presents the benchmark results, methodology, model comparisons, robustness analyses, reproducibility information, and limitations for the NEUROMOYO evaluation on African code-switched speech.

🚀 Live NEUROMOYO Demo

Live application: https://www.neuromoyo.app

The live NEUROMOYO application demonstrates the clinician-oriented voice intelligence workflow integrating speech recognition, clinical information extraction, and neurological voice-screening support.

Demo Access

A dedicated demo account can be used to access the application:

Email: staranye@gmail.com Password: anye1000

These credentials are provided specifically for evaluation of the NEUROMOYO demonstration application.

💻 GitHub Repository

Source code: https://github.com/ProKelly/neuromoyo.git

The GitHub repository contains the NEUROMOYO application source code and implementation of the clinician-oriented voice intelligence workflow.

📊 Evidence Package

This repository contains the reproducibility and evidence materials supporting the benchmark:

• —benchmark/all_results.csv — overall model results
• —benchmark/cmi_slices.csv — CMI-stratified results
• —benchmark/switch_slices.csv — switch-point-stratified results
• —benchmark/duration_slices.csv — duration-stratified results
• —benchmark/paired_sahara_comparisons.csv — paired model comparisons
• —benchmark/wer_bootstrap_ci.csv — bootstrap confidence intervals
• —benchmark/correlations.csv — correlation analyses
• —benchmark/evaluation_manifest.csv — evaluation manifest
• —benchmark/BENCHMARK_REPORT.md — detailed benchmark report
• —NEUROMOYO_Sahara_CodeSwitch_Africa_Challenge.ipynb — reproducible notebook

Additional supporting materials:

• —benchmark/ — CSV results and evaluation manifest
• —figures/ — benchmark figures
• —reproducibility/ — run configuration metadata
• —responsible_ai/ — responsible AI and inclusion note

Benchmark Scope

The evaluation uses 100 Pidgin utterances from the Intron Health AfriSwitch test set and compares:

• —Sahara
• —MMS-1B
• —Whisper Tiny
• —Whisper Base

Raw AfriSwitch audio is not redistributed in this repository. The repository contains aggregate benchmark evidence and reproducibility materials.

🧠 NEUROMOYO Voice Intelligence Workflow

The benchmark is connected to a broader clinician-oriented voice intelligence workflow:

Patient speech ↓ Audio quality assessment ↓ Sahara speech recognition ↓ Clinical information extraction ↓ Existing acoustic neurological screening pathway ↓ Evidence and provenance ↓ Clinician decision support

Sahara provides the speech recognition layer. It does not perform Parkinson's disease diagnosis.

The neurological screening component analyses acoustic speech characteristics independently from the transcript.

🛡️ Responsible AI

NEUROMOYO is designed as a clinical decision-support and screening system rather than an autonomous diagnostic system.

The implementation includes:

• —explicit consent before voice analysis
• —temporary audio processing
• —no redistribution of raw benchmark audio
• —provenance for extracted clinical information
• —evidence snippets supporting extracted findings
• —human clinician review
• —explicit limitations around clinical interpretation

The ASR benchmark should not be interpreted as clinical diagnostic validation.

🔍 Quick Navigation

Challenge

Sahara CodeSwitch Africa Main Challenge 2026

Project: NEUROMOYO Category: Health

NEUROMOYO turns natural African speech into structured clinical information and neurological voice-screening support for healthcare teams.