voxa-dub 1.0.1__tar.gz

voxa_dub-1.0.1/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Akshin Miranov
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

voxa_dub-1.0.1/NOTICE.md

@@ -0,0 +1,127 @@
+# Third-Party Notices
+
+**Voxa** is released under the [MIT License](LICENSE).
+
+Voxa **does not bundle or redistribute any model weights or third-party source code**.
+It orchestrates tools and models that you install and run yourself. The licenses of those
+components are their own, and some of them are **more restrictive than Voxa's MIT license**.
+
+Read this file before using Voxa commercially.
+
+---
+
+## 1. Quick answer: what is safe for commercial use?
+
+| Configuration | Commercially usable? | Why |
+|---|:---:|---|
+| `--tts piper` + `--translator ollama` | ✅ Yes | Fully offline, permissive licenses |
+| `--tts openai` + `--translator openai` | ✅ Yes | Paid API under OpenAI's commercial terms |
+| `--tts edge` (default) | ⚠️ Grey area | Uses an **unofficial** Microsoft endpoint (see §3) |
+| `--translator google` (default) | ⚠️ Grey area | Uses an **unofficial** Google endpoint (see §3) |
+| `--tts xtts` | ❌ **No** | XTTS-v2 model weights are **non-commercial only** (see §4) |
+
+> **The default configuration (`--tts edge --translator google`) is convenient for personal use
+> but is NOT recommended for a commercial product.** Prefer `openai`/`anthropic` translation and
+> `openai` or `piper` synthesis, or a self-hosted OpenAI-compatible endpoint.
+
+---
+
+## 2. Runtime dependencies
+
+Licenses below were read from the installed package metadata.
+
+### Required
+
+| Package | Role | License |
+|---|---|---|
+| [openai-whisper](https://github.com/openai/whisper) | Speech recognition | MIT |
+| [edge-tts](https://github.com/rany2/edge-tts) | Default TTS engine | **LGPL-3.0** |
+| [deep-translator](https://github.com/nidhaloff/deep-translator) | Google translation backend | MIT |
+| [openai](https://github.com/openai/openai-python) | OpenAI API client | Apache-2.0 |
+| [anthropic](https://github.com/anthropics/anthropic-sdk-python) | Anthropic API client | MIT |
+| [torch](https://github.com/pytorch/pytorch) / torchaudio | Inference runtime | BSD-3-Clause |
+| [soundfile](https://github.com/bastibe/python-soundfile) | Audio I/O | BSD-3-Clause (libsndfile: LGPL) |
+| [noisereduce](https://github.com/timsainb/noisereduce) | Source-audio denoising | MIT |
+| [numpy](https://numpy.org/) | Numerics | BSD-3-Clause |
+| [tqdm](https://github.com/tqdm/tqdm) | Progress bars | MPL-2.0 AND MIT |
+| [requests](https://requests.readthedocs.io/) | Ollama HTTP calls | Apache-2.0 |
+
+### Optional
+
+| Package | Enables | License |
+|---|---|---|
+| [piper-tts](https://github.com/rhasspy/piper) | `--tts piper` | MIT |
+| [coqui-tts](https://github.com/idiap/coqui-ai-TTS) | `--tts xtts` | MPL-2.0 (toolkit only) |
+| [faster-whisper](https://github.com/SYSTRAN/faster-whisper) | `--whisper-backend faster` | MIT |
+
+### ⚠️ Copyleft dependencies you should know about
+
+- **`edge-tts` is LGPL-3.0.** Importing it without modification is generally compatible with an
+  MIT application, provided users remain able to replace the library. Do not vendor a modified
+  copy without complying with the LGPL. To ship a fully permissive bundle, use `--tts openai`
+  or `--tts piper` and do not install `edge-tts`.
+
+> **Removed:** Voxa used to require `pysrt` (GPL-3.0) for a single call — reading back the
+> subtitle file it had just written itself. That call was replaced with a small built-in SubRip
+> parser, so **no GPL-licensed package is required to run Voxa.**
+
+---
+
+## 3. Unofficial service endpoints
+
+Two default backends talk to endpoints that are **not officially published APIs**:
+
+- **`edge-tts`** speaks to Microsoft's internal Edge "Read Aloud" speech service. It is not a
+  supported public API, has no SLA, and can break without notice — versions below `7.0` are
+  already rejected with HTTP 403. For a supported, commercially licensed equivalent, use
+  Azure Cognitive Services Speech.
+- **`deep-translator`'s Google backend** scrapes the public Google Translate web endpoint. It is
+  not the paid Google Cloud Translation API.
+
+Using them is your decision and your risk. For production or commercial work, prefer an official,
+paid API (OpenAI / Anthropic for translation; OpenAI TTS or Azure Speech for synthesis).
+
+---
+
+## 4. Models
+
+Voxa downloads models at runtime; it never ships weights.
+
+| Model | License | Commercial use |
+|---|---|---|
+| Whisper weights (OpenAI) | MIT | ✅ Allowed |
+| Piper voices | Per-voice (commonly MIT / CC-BY) | ✅ Check the individual voice |
+| **XTTS-v2 weights (Coqui)** | **Coqui Public Model License (CPML) 1.0** | ❌ **Non-commercial only** |
+
+### About XTTS
+
+Coqui Inc. **ceased operations in January 2024**. Consequences:
+
+1. The original `TTS` PyPI package is **unmaintained**. Voxa's optional extra installs the
+   community-maintained fork [`coqui-tts`](https://github.com/idiap/coqui-ai-TTS) instead.
+2. The XTTS-v2 **model weights remain under CPML, which permits non-commercial use only**, and
+   because the company no longer exists, **there is no party from whom a commercial license can
+   be obtained.**
+
+If you need voice cloning for a commercial product, XTTS is not a viable path. Consider an
+MIT-licensed alternative such as [Chatterbox](https://github.com/resemble-ai/chatterbox), served
+behind an OpenAI-compatible endpoint.
+
+---
+
+## 5. External APIs
+
+`--translator openai`, `--translator anthropic` and `--tts openai` send your transcript and/or
+text to third-party services under their own terms of service and pricing. Voxa never transmits
+your API keys anywhere except to the corresponding provider.
+
+For a fully offline pipeline that sends nothing over the network, use:
+
+```bash
+voxa video.mp4 --translator ollama --tts piper
+```
+
+---
+
+*If you believe any license above is stated incorrectly, please open an issue — accuracy here
+matters more than convenience.*