Sisi Lola Voice - Nigerian TTS Profile
Model Details
Model Description
Sisi Lola Voice is a Nigerian Text-to-Speech voice profile designed for generating natural-sounding Nigerian English speech. This profile configures TTS engines (EdgeTTS, XTTS) with Nigerian voice characteristics including accent, intonation, and cultural speech patterns.
- Developed by: BAMG Studio
- Model type: TTS Voice Profile / Configuration
- Language(s): English (Nigerian accent), Nigerian Pidgin, Yoruba
- License: Apache 2.0
- Compatible with: EdgeTTS (en-NG voices), XTTS-v2, YarnGPT
- Repository: BAMG-Studio/sisi-lola-project
Model Sources
Uses
Direct Use
This voice profile is designed to be used with TTS engines to generate Nigerian-accented speech for virtual assistants, chatbots, and audio content.
Primary use cases:
- Voice synthesis for Sisi Lola chatbot
- Nigerian English audio content generation
- Accessibility features with Nigerian accent
- Educational content with authentic Nigerian pronunciation
Supported TTS Engines
| Engine | Voice | Language |
|---|
| EdgeTTS | en-NG-EzinneNeural | Nigerian English (Female) |
| EdgeTTS | en-NG-AbeoNeural | Nigerian English (Male) |
| XTTS-v2 | Custom cloning | Multi-language |
| YarnGPT | Nigerian voices | Yoruba, Pidgin, Igbo, Hausa |
Out-of-Scope Use
This voice profile should NOT be used for:
- Impersonating real individuals
- Generating misleading audio content
- Deepfake or deceptive purposes
- Any illegal activities
How to Get Started
With EdgeTTS (Recommended)
```python
import edge_tts
import asyncio
async def speak(text, output_file="output.mp3"):
communicate = edge_tts.Communicate(
text,
voice="en-NG-EzinneNeural", # Nigerian female voice
rate="+5%"
)
await communicate.save(output_file)
print(f"Audio saved to {output_file}")
Example
asyncio.run(speak("How you dey? Welcome to Lagos!"))
```
With Hybrid Voice Stack
```python
from hybrid_voice_stack import HybridVoiceStack
Initialize with Nigerian language support
voice = HybridVoiceStack()
Synthesize with automatic language detection
audio = voice.synthesize("Ẹ kú àárọ̀! How you dey today?")
audio.save("greeting.mp3")
```
Technical Specifications
Voice Profile Configuration
```json
{
"name": "Sisi Lola",
"accent": "Nigerian",
"region": "Lagos",
"primary_language": "en-NG",
"secondary_languages": ["pcm", "yo"],
"voice_characteristics": {
"gender": "female",
"age_range": "25-35",
"tone": "warm",
"style": "conversational"
}
}
```
Language Routing
The voice profile supports automatic language detection and routing:
| Tag | Language | Engine |
|---|
| [EN] | Nigerian English | EdgeTTS |
| [NP] | Nigerian Pidgin | YarnGPT/EdgeTTS |
| [YO] | Yoruba | YarnGPT |
| [IG] | Igbo | YarnGPT |
| [HA] | Hausa | YarnGPT |
Evaluation
Voice Quality
- Natural Nigerian English intonation
- Clear pronunciation
- Appropriate pacing for conversational use
- Warm, friendly tone
Limitations
- EdgeTTS requires internet connection
- Limited Yoruba/Pidgin vocabulary in fallback mode
- May not capture all regional Nigerian accents
Model Card Contact
Sisi Lola - Your friendly Lagos virtual host 🇳🇬
