Datasetpublic

ODYSSEYAILABS/odyssey-v0.5.1-16h-eval-preview-public

🇿🇦 Odyssey V0.5.1 — 16H Evaluation Preview (Code-Switched ASR) Odyssey AI Labs presents a 16-hour Enterprise Evaluation Preview of the V0.5 corpus. This release is designed to surface the hardest South African ASR realities directly: native urban code-switching, multi-speaker turn-taking, and overlapping conversational speech. A structured public preview of the Odyssey SA Voice Corpus, designed for researchers, data buyers, and speech teams evaluating multilingual South… See the full description on the dataset page: https://huggingface.co/datasets/ODYSSEYAILABS/odyssey-v0.5.1-16h-eval-preview-public.

sourceHugging Faceotherupdated 5mo agoView on Hugging Face

0likes58downloads

Dataset Card

🇿🇦 Odyssey V0.5.1 — 16H Evaluation Preview (Code-Switched ASR)

Odyssey AI Labs presents a 16-hour Enterprise Evaluation Preview of the V0.5 corpus. This release is designed to surface the hardest South African ASR realities directly: native urban code-switching, multi-speaker turn-taking, and overlapping conversational speech.

A structured public preview of the Odyssey SA Voice Corpus, designed for researchers, data buyers, and speech teams evaluating multilingual South African conversational data.

This repository is a transcript-and-metadata release. It is intended to demonstrate data quality, schema design, diarization structure, and real-world linguistic complexity before commercial engagement.

Raw audio is not distributed in this repository.

What This Preview Contains

—171 conversational clips
—16.15 hours of source material
—Train / Validation / Test split files
—Public transcript text
—Speaker-count and code-switch indicators
—QA summary fields
—Sidecar metadata, transcript, and QA JSON files
—Preserved source lineage IDs

Why It Matters

Many speech datasets perform well in controlled settings but fail in real environments.

This preview focuses on real South African conditions such as:

—natural code-switching
—multi-speaker conversations
—turn-taking dynamics
—overlapping speech
—multilingual utterances
—informal spoken language

These are the conditions production systems must handle.

Data Structure

Main split files:

—data/train.jsonl
—data/validation.jsonl
—data/test.jsonl

Each row includes fields such as:

—id
—batch_id
—clip_number
—duration_seconds
—primary_language
—secondary_language
—speaker_count
—code_switch
—text

Additional sidecars:

—sidecars/metadata/
—sidecars/transcripts/
—sidecars/qa_flags/

Intended Use

Allowed:

—dataset inspection
—schema review
—ingestion testing
—buyer diligence
—research evaluation

Not allowed:

—commercial deployment
—model training
—fine-tuning
—redistribution
—derivative dataset release

Commercial use requires a separate agreement with Odyssey AI Labs.

Companion Gated Audio Repository

The companion audio repository contains gated evaluation audio for approved users:

ODYSSEYAILABS/odyssey-v0.5.1-16h-eval-audio-gated

Access requests are reviewed manually.

About Odyssey AI Labs

Odyssey AI Labs builds high-quality African data infrastructure for modern AI systems, beginning with multilingual South African voice data.