The largest open-source humanized voice library — 4,065 voice samples across 44 languages, free forever under CC-BY-SA 4.0.

We collected raw AI/synthetic voice samples from 10 open-source datasets, then humanized every single one through our proprietary voice processing pipeline. The original synthetic artifacts — robotic tone, unnatural cadence, flat prosody — are replaced with natural-sounding speech while preserving each voice's unique timbre, pitch, and character.

The result: a massive library of natural-sounding reference voices that anyone can use for TTS, voice cloning, dubbing, audiobooks, and creative projects.

voices/
├── english/
│   ├── male/
│   │   ├── english_male_open_swara_001.wav
│   │   └── ...
│   ├── female/
│   └── unknown/
├── german/
│   ├── male/
│   │   ├── german_male_open_swara_001.wav          ← speaking German
│   │   ├── german_male_english_open_swara_001.wav  ← speaking English with German accent
│   ├── female/
│   └── unknown/
├── french/
│   └── ...
└── ... (44 languages)

• {language}_{gender}_open_swara_{NNN}.wav — voice speaking in its native language
• {language}_{gender}_english_open_swara_{NNN}.wav — voice speaking English with native accent

Arabic, Bulgarian, Catalan, Chinese, Czech, Danish, Dutch, English, Finnish, French, Georgian, German, Greek, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Kazakh, Korean, Lao, Latvian, Luxembourgish, Malayalam, Nepali, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swahili, Swedish, Telugu, Turkish, Twi, Ukrainian, Vietnamese, Welsh

These voices are ideal as reference samples for:

• Text-to-Speech (TTS) voice cloning
• Voice conversion systems
• Multilingual dubbing and localization
• Audiobook narration prototyping
• Creative and research projects

# Search for German female voices
python scripts/browse_catalog.py --language german --gender female

# List all languages
python scripts/browse_catalog.py --list-languages

# Search by keyword
python scripts/browse_catalog.py --search "english_accent"

CC-BY-SA 4.0 — free for commercial and non-commercial use with attribution.

See LICENSE for full terms.

When using Open Swara voices, please credit:

Voices from Open Swara.

See ATTRIBUTION.md for source dataset credits.