About and Licenses

Models

Product names appear alongside the original model names and creators. Availability depends on your installation; this inventory covers supported models. Model weights and source code may have different licenses.

• EMA Lightning — Canberk Aslan

  Emira· Apache-2.0

  SintelSES adaptation: ONNX model conversion and inference engine.

• Ahenk Cep — Ahenk / Ozgur Guler

  Ahen· CC-BY-4.0 / Apache-2.0

  SintelSES adaptation: FP32 ONNX inference engine, Turkish text normalization and SSML. Model weights use CC-BY-4.0; the EMA base and source code use Apache-2.0.

• Supertonic 3 — Supertone Inc.

  Sonik· OpenRAIL-M / MIT

  SintelSES adaptation: inference engine and application integration. Model weights use OpenRAIL-M; source code uses MIT.

• Pocket TTS Turkish — Wite Tech / Kyutai

  Piko · tr· CC-BY-4.0

  SintelSES adaptation: FP32 ONNX conversion, explicit streaming state and cached built-in voices.

• Pocket TTS — Kyutai

  Piko · en, fr, de, pt, it, es, nl· CC-BY-4.0

  SintelSES adaptation: FP32 ONNX conversion, explicit streaming state and cached built-in voices.

• Drifting TTS / Vocos — Vyvo, kadirnar, charactr / gemelo-ai

  Duru· CC-BY-4.0 / MIT

  SintelSES adaptation: v3.1 ONNX / Vocos-ft conversion, inference engine and Turkish text normalization. Model weights use CC-BY-4.0; code and Vocos use MIT.

• mihu Turkish STT — Hunters AI

  Mira· mihu Community License 1.0

  Powered by mihu

• Nemotron 3.5 ASR — NVIDIA / csukuangfj2

  Nemo· OpenMDW-1.1

• Duyu-small — Kaan Aydinli

  Duyum· Apache-2.0

  SintelSES adaptation: inference engine and application integration.

• Turkish ONNX aligner — Parseh / mpoyraz / Meta

  CC-BY-4.0

  Parseh adaptation: ONNX conversion and int8 quantization of the mpoyraz / Meta-based model, without retraining.

• ECAPA-TDNN — SpeechBrain

  Apache-2.0

  Used to extract speaker embeddings in the optional voice cloning tools.