MANK TTS · TURKISH VOICE TECHNOLOGY
Natural Turkish speech, built for production.
Enterprise-grade Turkish speech infrastructure by Mank.
Mank TTS is designed for enterprise voice experiences that require natural pronunciation, reliable production quality, real-time streaming and optional custom voice capabilities. Hear each model in your own scenario and we will pick the right fit together.
Avaz 2
Stable for productionFor predictable quality and always-on operations.
- Primary use:
- Call greeting and notification flows
Avaz 3
More natural expressionStronger prosody and a more fluent speaking experience.
- Primary use:
- Voice assistants and customer-facing dialogue
Avaz 3.5
Coming soonEven more natural and expressive speech synthesis.
- Primary use:
- Brand voice and highly expressive experiences
Payment reminder call
Transcript
Hello, we wanted to remind you that your credit card payment due date is approaching. We can review the payment plan options together.
Capabilities of the selected model — Avaz 2
Capabilities depend on the model; for unapproved details we suggest a conversation.
Streaming
AvailableFor real-time synthesis and live conversational streams.
Custom pronunciation
Custom projectPronunciation control for internal terms, brand names and abbreviations.
Deployment options
AvailableOn-premise, private cloud or Mank infrastructure located in Türkiye.
Language support
Coming soonTurkish is the primary language; other languages are assessed per project.
Voice cloning
Let's discuss the detailsConfigurable voice cloning for organization-specific voice experiences.
Brand voice
Let's discuss the detailsOne consistent corporate voice across every channel.
Choose the right model for your needs
| Avaz 2 | Avaz 3 | Avaz 3.5 | |
|---|---|---|---|
| Primary use | Call greeting and notification flows | Voice assistants and customer-facing dialogue | Brand voice and highly expressive experiences |
| Naturalness | Co-one evaluation score: 25/27 | Co-one evaluation score: 27/27 | Benchmark data will be added soon. |
| Production readiness | In production | In production | Coming soon |
| Real-time streaming | Available | Available | Available |
| Voice cloning | Available | Yes | Coming soon |
| Recommended deployment | On-premise, private cloud or Mank infrastructure | On-premise, private cloud or Mank infrastructure | On-premise, private cloud or Mank infrastructure |
| Audio demo availability | Yes | Yes | Demo coming soon |
Approved facts
- Built for Turkish
- Optimized for enterprise operations
- Real-time usage options
- On-premise and private deployment options
Technical details
The Avaz family is built on neural speech-synthesis layers adapted to Turkish phonology and stress patterns. Architecture details are shared in a technical session.
Models are developed with large-scale pretraining followed by domain and customer-specific adaptation. Scale and data-source figures are published only once verified.
On-premise, private cloud or Mank infrastructure located in Türkiye. The choice depends on data ownership, network policy and operating model.
Where audio is processed and stored depends on the chosen deployment model. Certification and compliance statements are listed only once approved.
A pronunciation lexicon can be defined for internal terms, product and brand names; updates require no re-recording.
Benchmark data will be added soon. Evaluation is run on your own scripts and scenarios.
Let's choose the right voice for your operation.
Share your scenario and we will assess the right TTS model and demo flow together.