jml2026/multilingual-accent-speech

🎙️ Silencio Network: Voice AI Sample Dataset

<p align="left"> <img src="https://cdn-uploads.huggingface.co/production/uploads/69162b50b89e7abe20de4b5a/LWhs4p2lPFcyiVsP0tluu.png" width="40%"> </p>

  

📊 This is a sample. The full Silencio corpus contains 100,000+ hours across 170+ countries and 100+ languages. 📧 Contact: sofia@silencioai.com for custom datasets, bulk licensing, or specific language requests.

🌍 Why Silencio Data?

Silencio data is collected in the wild from a massive, opt-in community (2M+ contributors across 180+ countries), giving you:

• —✅ Real-world accents, dialects, devices, and environments that lab or scraped datasets don't capture
• —✅ Explicit, traceable consent — every recording tied to verified opt-in (GDPR/CCPA compliant)
• —✅ Privacy-first pipelines — anonymized, PII hashed, reduced legal risk for enterprise
• —✅ Rapid scaling into hard-to-source languages and niches

📊 Full Data Availability

Custom requests? We can source virtually any language, accent, or demographic at scale.

🎯 This Sample Dataset

This sample covers 43 language–region configs across 20+ languages demonstrating our data quality:

Accented English

African Languages

Other Languages

Medical Domain

Conversational

Each row contains 3 audio files (individual speaker, full meeting single-channel, full meeting multi-channel) plus word-level meeting transcripts with speaker turns and timestamps.

🚀 Quick Start

from datasets import load_dataset

# Load Nigerian English samples
ds = load_dataset("SilencioNetwork/complete-voiceai-speech-dataset", "english_nigeria")

# Access different speech types
free_speech = ds['free_speech']
keywords = ds['keywords']
monologues = ds['monologues']

# Process a sample
for sample in free_speech:
    audio = sample['audio']
    transcript = sample['transcript']
    speaker_id = sample['speaker_id']
    emotion = sample['emotions']
    print(f"Speaker {speaker_id}: {transcript[:50]}...")

🎤 Speech Types

📋 Rich Metadata

Every recording includes:

• —Speaker demographics: gender, ethnicity, occupation, birth year, birth place
• —Linguistic info: mother tongue, dialect, language proficiency data
• —Recording context: device, OS, browser, location, background noise
• —Content: script, Whisper-generated transcript, emotion labels
• —Technical: duration, 48kHz WAV format

🎯 Ideal For

• —🗣️ ASR training — accent-robust speech recognition
• —🔊 TTS development — diverse voice synthesis
• —😊 Emotion recognition — labeled emotional speech
• —🌍 Multilingual models — 100+ languages available
• —🎙️ Speaker verification — unique speaker embeddings
• —📊 Benchmarking — real-world robustness testing

⚖️ License

CC BY-NC 4.0 — Free for research and non-commercial use.

For commercial licensing, contact sofia@silencioai.com.

📧 Get More Data

Need the full corpus? This sample represents <0.01% of our available data.

📧 Email: sofia@silencioai.com

🌐 Website: silencioai.com

Citation

@dataset{silencio_network_speech_2025,
    title        = {Silencio Network Voice AI Speech Corpus},
    author       = {Silencio Network},
    year         = {2025},
    publisher    = {Hugging Face},
    license      = {CC BY-NC 4.0}
}