Open-source release

.gitattributes

@@ -0,0 +1,38 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer-multi_bpe16384_merged_extended_58M.json filter=lfs diff=lfs merge=lfs -text
+v090_g_01105000 filter=lfs diff=lfs merge=lfs -text
+assets/DiscreteDiffusionSchema.png filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,201 @@
+  Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2025 PlayHT, Inc.
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -0,0 +1,86 @@
+---
+title: PlayDiffusion
+emoji: 🎨
+colorFrom: green
+colorTo: yellow
+license: apache-2.0
+language:
+- en
+---
+
+# PlayDiffusion
+
+## Introduction
+
+
+Autoregressive transformer models have proven highly effective for synthesizing speech from text. However, they face a significant limitation: **modifying portions of the generated audio —known as inpainting— or removing them without leaving discontinuity artifacts is beyond their standard capabilities**. Thus different approaches are needed for more general speech editing tools. Consider the sentence:
+
+> *“The answer is out there, Neo. Go grab it!”*
+
+Now suppose you want to change 🔴 "Neo" to 🟢 "Trinity" *after* generation. With traditional AR models, your options are limited.
+- **Regenerate the entire sentence**, which is computationally expensive and often leads to variations in prosody or speech rhythm.
+- **Replace just the word “Neo”**, which results in noticeable artifacts or mismatches at word boundaries.
+- **Regenerate from a midway point**, e.g., from “Trinity. Go grab it!”, but this has the potential of changing the prosody of the unedited part ("Go grab it"), creating unwanted variations in the speech rhythm.
+
+All of these approaches compromise both the coherence and naturalness of the audio. Instead, for these type of speech editing tools, non-autoregressive approaches are favorable [[1](https://arxiv.org/pdf/2401.04577), [2](https://arxiv.org/pdf/2305.09636)].
+
+
+## PlayDiffusion Model
+
+At [Play.AI](http://play.ai/), we’ve addressed this problem with a **novel diffusion-based approach for audio speech editing**. Here’s how it works:
+
+1. First, we encode an audio sequence into a discrete space, converting the waveform into a more compact representation. Each unit in this representation is called a **token**. This process works with both real speech and audio generated by Text-to-Speech models.
+2. When a segment needs to be modified, we **mask** that portion of the audio.
+3. A **diffusion model**, conditioned on the updated text is used to **denoise** the masked region.
+    1. The surrounding context is preserved seamlessly, ensuring smooth transitions and consistent speaker characteristics.
+4. The resulting output token sequence is then transformed back to a speech waveform using our BigVGAN decoder model.
+
+By using a **non-autoregressive diffusion model**, we can better maintain context at the edit boundaries—resulting in high-quality, coherent audio edits. This marks a significant step forward in audio editing capabilities and paves the way for dynamic, fine-grained speech modification.
+
+The full process is outlined in Figure 1.
+
+![Image](https://github.com/user-attachments/assets/16f783cd-c9c5-4c60-aabc-c57dc4b1f894)
+> **Figure 1.** PlayDiffusion Model. 1) The input audio containing the speech “The answer is out there Neo. Go grab it!” is encoded to discrete audio tokens. 2) The tokens that corresponds to the speech targeted for editing is masked. Here we mask the tokens for “Neo”. 3) Given the updated text and the full input token sequence (i.e., masked and unmasked) the PlayDiffusion model generates the edited output sequence. 4) The sequence is transformed to a waveform by our BigVGAN, conditioned on the speaker embedding extracted from the original clip.
+
+## Training
+
+We started with a pre-trained decoder-only text-transformer architecture and introduced crucial modifications tailored specifically for audio generation:
+
+1. **Non-Causal Masking:**
+Unlike standard decoder-only LLMs, like GPT, which employ causal attention (allowing tokens only to attend to previous ones), our modified Llama implementation uses non-causal attention heads. This allows the model to simultaneously leverage past, present, and future tokens.
+2. **Custom Tokenizer and Embedding Reduction:**
+To optimize efficiency, particularly for English speech synthesis, we used a custom Byte Pair Encoding (BPE) tokenizer consisting of only 10,000 text tokens. This drastically decreases the size of the embedding table, significantly boosting computational speed without compromising audio quality.
+3. **Speaker Conditioning:**
+Our model incorporates speaker conditioning derived from a pre-trained embedding model $e(w):\R^t \to \R^k$, which maps waveforms of varying length $t$ to fixed-size vectors of dimension $k$. This captures essential speaker characteristics, ensuring consistent voice identity across synthesized or edited audio segments.
+
+During **training**, like MaskGCT [[3]](https://arxiv.org/pdf/2409.00750), we randomly mask a percentage of audio tokens. The model learns to accurately predict these masked tokens based on context provided by the speaker embeddings, textual input, and remaining unmasked audio tokens. This method effectively trains the model to handle partial or complete audio masking scenarios. Given text samples at different time-steps as $x_t$ and text condition $\mathbf{C}$, we model the loss as:
+
+$$
+\mathcal{L}_{\text{mask}} = - \sum _{\mathcal{X} \in \mathcal{D}, t \in \[0,T]} \sum _{i=1}^{N} m _{t,i} \cdot \log p{\theta}(x_i | \mathbf{X}_t, \mathbf{C}).
+$$
+
+where the mask $m_{t,i}$ is defined as:
+
+$$
+m_{t,i} = \begin{cases}
+1, & \text{if token } i \text{ is masked} \\
+0, & \text{otherwise}
+\end{cases}
+$$
+
+During **inference**, decoding occurs iteratively, starting with a **fully masked token** sequence. The decoding process is conducted over multiple steps, designated as T:
+
+- **Step-wise Decoding Process:**
+    1. **Preliminary Prediction:** At each iteration, the model generates an initial prediction, X̂₀,  conditioned on the current masked audio and textual input.
+    2. **Confidence Scoring:** Tokens receive a confidence score based on the model’s prediction. Newly predicted (previously masked) tokens are assigned a confidence equivalent to their predicted probabilities, while tokens previously determined remain unaltered with a confidence score of 1.
+    3. **Adaptive Remasking:** Utilizing a progressively decreasing schedule, $\gamma$, we select a specific number of lowest-confidence tokens for remasking in subsequent iterations. The number of tokens to remask reduces progressively with each iteration, focusing the model’s refinement efforts on areas of highest uncertainty. See MaskGCT [[3]](https://arxiv.org/pdf/2409.00750) for further details.
+
+This iterative decoding process continues until all steps are complete, gradually refining token predictions and resulting in coherent, high-quality audio outputs.
+
+## **Reference:**
+
+[1] Alon Ziv, Itai Gat, Gael Le Lan, Tal Remez, Felix Kreuk, Alexandre Défossez, Jade Copet, Gabriel Synnaeve, and Yossi Adi. 2024. Masked audio generation using a single non-autoregressive transformer. arXiv preprint [arXiv:2401.04577](https://arxiv.org/pdf/2401.04577).
+
+[2] Zalán Borsos, Matt Sharifi, Damien Vincent, Eugene Kharitonov, Neil Zeghidour, and Marco Tagliasacchi. 2023. SoundStorm: Efficient parallel audio generation. arXiv preprint [arXiv:2305.09636](https://arxiv.org/pdf/2305.09636).
+
+[3] Yuancheng Wang, Haoyue Zhan, Liwei Liu, Ruihong Zeng, Haotian Guo, Jiachen Zheng, Qiang Zhang, Xueyao Zhang, Shunsi Zhang, and Zhizheng Wu. 2025. Maskgct: Zero-shot text-to-speech with masked generative codec transformer. In ICLR. [Arxiv link](https://arxiv.org/pdf/2409.00750).

kmeans_10k.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd92ff1a89a67e6a8a5c3881dc479065717111e7356490842c92cd9aa54aeaf5
+size 51200128

last_250k_fixed.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddd3b19e2099d27356ed89492af9ea23fcb8d76b8f717574c416795f0828dd12
+size 4439103514

tokenizer-multi_bpe16384_merged_extended_58M.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a717a09fab1e1bbfdbe1f5a3a2cd369391de3b6afd240a4b5522f0110157c10e
+size 1140805

v090_g_01105000

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:264a236d6dbc50cf95e21df93c44eaf73dd2d19b0a02fa8b5e3ad62d3f86a36a
+size 1670315646

voice_encoder_1992000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e741b032861aa8f68b16c0b8bb7a5add6b1387be5fde29d9b8d8d81cc2a27f22
+size 304493771

xlsr2_1b_v2_custom.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c28eaa338f11059581c9c1a8441d298c1510d8036bfa70f9a720dcd837ce8e6f
+size 3862369404