Licences & credits
sōtem models are built on these open foundations, under the licences their authors published. Each is pinned to an exact revision and runs only on Mondaily's own servers. sōtem Khaleeji is ours, trained on sōtem Data Lab recordings.
sōtem models and their open foundations
| sōtem model | Foundation | Author / copyright holder | Licence | Link | Required notice |
|---|---|---|---|---|---|
| sōtem V1 | SILMA TTS | SILMA AI | Apache-2.0 | Source | Licence text linked. Upstream NOTICE kept where the project provides one. |
| Chatterbox Multilingual | Resemble AI | MIT | Source | Copyright (c) 2025 Resemble AI. Permission notice: MIT licence (linked). | |
| Vocos (mel, 24 kHz) | Charactr | MIT | Source | Copyright (c) Charactr. Permission notice: MIT licence (linked). | |
| sōtem Scribe | Cohere Transcribe Arabic | Cohere Labs | Apache-2.0 | Source | Licence text linked. Upstream NOTICE kept where the project provides one. |
| Qwen3-ASR | Qwen team | Apache-2.0 | Source | Licence text linked. Upstream NOTICE kept where the project provides one. | |
| pyannote community-1 | pyannoteAI | CC-BY-4.0 | Source | speaker-diarization-community-1 by pyannoteAI, licensed under CC BY 4.0. Not modified — used as a component. | |
| NVIDIA FastConformer Arabic | NVIDIA | CC-BY-4.0 | Source | stt_ar_fastconformer_hybrid_large_pcd_v1.0 by NVIDIA, licensed under CC BY 4.0. Not modified — used as a component. | |
| Whisper large-v3-turbo | OpenAI | MIT | Source | Copyright (c) 2022 OpenAI. Permission notice: MIT licence (linked). | |
| wav2vec2 XLSR-53 Arabic | Jonatas Grosman | Apache-2.0 | Source | Licence text linked. Upstream NOTICE kept where the project provides one. | |
| sōtem Echo | Chatterbox VC | Resemble AI | MIT | Source | Copyright (c) 2025 Resemble AI. Permission notice: MIT licence (linked). |
| OpenVoice v2 | MyShell | MIT | Source | Copyright (c) MyShell. Permission notice: MIT licence (linked). | |
| sōtem Dub | MADLAD-400 | Apache-2.0 | Source | Licence text linked. Upstream NOTICE kept where the project provides one. | |
| sōtem Sound | MOSS-SoundEffect | OpenMOSS | Apache-2.0 | Source | Licence text linked. Upstream NOTICE kept where the project provides one. |
| MOSS-Audio-Tokenizer | OpenMOSS | Apache-2.0 | Source | Licence text linked. Upstream NOTICE kept where the project provides one. | |
| sōtem Muse | ACE-Step 1.5 | ACE Studio & StepFun | MIT | Source | Copyright (c) ACE-Step. Permission notice: MIT licence (linked). |
| sōtem Design | Being built — no foundation chosen yet. | ||||
| sōtem Clean | DeepFilterNet3 | Hendrik Schröter | MIT | Source | Copyright (c) 2021 Hendrik Schröter. Permission notice: MIT licence (linked). |
| sōtem Guard | AudioSeal | Meta | MIT | Source | Copyright (c) Meta Platforms, Inc. and affiliates. Permission notice: MIT licence (linked). |
| AASIST | NAVER Corp. (CLOVA AI) | MIT | Source | Copyright (c) 2021-present NAVER Corp.. Permission notice: MIT licence (linked). | |
| sōtem Agents | Built on sōtem Scribe, sōtem and sōtem V1 — no separate foundation. | ||||
| sōtem Khaleeji | Our own model — trained on sōtem Data Lab recordings. | ||||
Reference recordings behind the example clips
The example clips on the sōtem Voice pages were generated by sōtem V1 using a reference recording from an openly licensed dataset. The person in that recording has no connection with sōtem and does not endorse it.
| Dataset | Licence | Attribution | Modifications |
|---|---|---|---|
| MBZUAI/ClArTTS (LibriVox public-domain recording) | CC BY 4.0 | Kulkarni et al., ClArTTS, Interspeech 2023 | Modified: trimmed, resampled; used as a voice reference. |