SilencioNetwork/global-french-speech

🌍 Global French Speech Dataset

🎯 Overview

The Global French Speech Dataset provides high-quality, real-world speech recordings from native French speakers across 30+ Francophone countries and regions. This dataset includes 50 audio files from France and Canada, representing the diversity of French language across continents.

French is spoken by 300+ million people globally across Europe, Africa, Americas, and Oceania. This dataset captures accent diversity from 2 major French variants with comprehensive off-the-shelf inventory available for 30+ regions.

Key Features

✅ 2 major French variants with samples - Metropolitan French (France) and Canadian French
✅ 50 audio recordings - Native French speakers
✅ 30+ Francophone regions available OTS - Africa, Europe, Americas
✅ Rich demographic metadata - Gender, age, occupation, location, dialect
✅ Real-world acoustic conditions - Natural environments
✅ 156,000+ OTS recordings - 1,600+ hours available commercially

🗂️ This is a Sample Dataset

These 50 recordings represent a sample of Silencio's capabilities. Full off-the-shelf inventory available:

Country/Region	OTS Recordings	OTS Hours	In This Sample?
France	33,309	517 hours	✅ 25 files
Senegal	11,272	356 hours	❌ Contact
Benin	16,385	276 hours	❌ Contact
Switzerland	1,732	138 hours	❌ Contact
Burkina Faso	6,490	100 hours	❌ Contact
Tunisia	1,503	86 hours	❌ Contact
Algeria	3,761	80 hours	❌ Contact
Andorra	892	68 hours	❌ Contact
Madagascar	4,690	66 hours	❌ Contact
Cameroon	6,206	66 hours	❌ Contact
Togo	8,723	62 hours	❌ Contact
Morocco	4,358	59 hours	❌ Contact
Nigeria	26,648	34 hours	❌ Contact
Canada	1,715	16 hours	✅ 25 files
20+ more regions	30,000+	200+ hours	❌ Contact
TOTAL	156,000+	1,600+ hours	50 files (~32 min)

Sample = 0.03% of available inventory (updated: March 30, 2026)

Geographic Coverage

Europe:

• France (Metropolitan French) - ✅ Samples available
• Switzerland, Belgium, Andorra, Monaco - OTS available

Africa (Francophone):

• West Africa: Senegal, Benin, Burkina Faso, Togo, Mali, Niger, Guinea, Côte d'Ivoire
• Central Africa: Cameroon, Congo, Gabon, Central African Republic
• North Africa: Algeria, Tunisia, Morocco
• East Africa: Madagascar, Comoros, Rwanda, Burundi

Americas:

• Canada (Quebec, New Brunswick) - ✅ Samples available
• Haiti, French Guiana, Caribbean territories

Silencio's Complete OTS Catalog:

• 📊 156,000+ French recordings from 30+ countries
• ⏱️ 1,600+ hours of French speech data
• 🌍 Every continent with French speakers represented
• ✅ Immediate commercial licensing available

Contact: sofia@silencioai.com for full catalog and pricing.

📊 Dataset Statistics

Metric	Value
Total Audio Files	50
French Variants	2
Total Speakers	40+ unique speakers
Audio Format	WAV (16-bit PCM)
Sample Rate	16 kHz / 44.1 kHz
Total Duration	~32 minutes
Geographic Coverage	France, Canada

Variant Distribution

Variant	Files	Region	Notes
Metropolitan French	25	France	Standard French, European accent
Canadian French	25	Canada	Quebec/Canadian accent, distinct from European
Total	50	2 countries	Native speakers

🎯 Use Cases

• Accent-Robust ASR: Train French speech recognition across regional variants
• Dialect Identification: Distinguish European vs Canadian vs African French
• TTS Development: Multi-accent French text-to-speech
• Linguistic Research: Study phonetic variation in Francophone world
• Model Evaluation: Test fairness across French-speaking demographics

📁 Dataset Structure

global-french-speech/
├── french_global/          # Metropolitan French (France)
│   └── free_speech/
│       ├── data/
│       │   ├── audio_*.wav
│       └── metadata.csv
└── french_canada/          # Canadian French (Quebec)
    └── free_speech/
        ├── data/
        │   ├── audio_*.wav
        └── metadata.csv

🚀 Getting Started

from datasets import load_dataset

# Load entire dataset
dataset = load_dataset("SilencioNetwork/global-french-speech")

# Load specific variant
french_france = load_dataset(
    "SilencioNetwork/global-french-speech",
    name="french_global"
)

canadian_french = load_dataset(
    "SilencioNetwork/global-french-speech",
    name="french_canada"
)

⚖️ License & Usage

License: CC BY-NC 4.0

✅ Research, academic, educational, non-commercial use
❌ Commercial products/services (requires licensing)

Commercial licensing: sofia@silencioai.com

🏢 About Silencio

Silencio provides scaled sourcing of real-world speech data. With 2M+ contributors across 180+ countries, we specialize in global language coverage including comprehensive Francophone dialect diversity.

Learn more: silencioai.com

📚 Citation

@dataset{silencio_global_french_2026,
  title={Global French Speech Dataset},
  author={Silencio Network},
  year={2026},
  publisher={HuggingFace},
  url={https://huggingface.co/datasets/SilencioNetwork/global-french-speech}
}

🤝 Contact

Email: sofia@silencioai.com
HuggingFace: Discussion Forum

---

Built by Silencio | Licensed under CC BY-NC 4.0

Tags: #French #Francophone #GlobalFrench #CanadianFrench #AfricanFrench #ASR #TTS #VoiceAI