danyalmalik commited on
Commit
54eb255
·
verified ·
1 Parent(s): 31fe5b7

Upload Chatterbox-ONNX-WebGPU

Browse files
Files changed (4) hide show
  1. LICENSE +21 -0
  2. README.md +45 -0
  3. onnx/conditional_decoder.onnx +3 -0
  4. voices/default.bin +3 -0
LICENSE ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ MIT License
2
+
3
+ Copyright (c) 2025 Resemble AI
4
+
5
+ Permission is hereby granted, free of charge, to any person obtaining a copy
6
+ of this software and associated documentation files (the "Software"), to deal
7
+ in the Software without restriction, including without limitation the rights
8
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9
+ copies of the Software, and to permit persons to whom the Software is
10
+ furnished to do so, subject to the following conditions:
11
+
12
+ The above copyright notice and this permission notice shall be included in all
13
+ copies or substantial portions of the Software.
14
+
15
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21
+ SOFTWARE.
README.md ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ base_model:
4
+ - ResembleAI/chatterbox
5
+ tags:
6
+ - onnx
7
+ - webgpu
8
+ - text-to-speech
9
+ - quick-inference
10
+ ---
11
+
12
+ # Chatterbox-ONNX-WebGPU
13
+
14
+ [ResembleAI/chatterbox](https://huggingface.co/ResembleAI/chatterbox) (English) for running in a web browser with ONNX
15
+ Runtime Web's WebGPU backend, as used by Quick Inference: a smaller speech decoder and a precomputed default voice.
16
+ The language model and token embeddings are onnx-community's, used unchanged from
17
+ [onnx-community/chatterbox-ONNX](https://huggingface.co/onnx-community/chatterbox-ONNX) (revision `3cab09af`):
18
+ `language_model_q4`, `embed_tokens` and `tokenizer.json`.
19
+
20
+ ## Changes from the original
21
+
22
+ - `onnx/conditional_decoder.onnx`: onnx-community's float32 `conditional_decoder.onnx` quantized without calibration
23
+ (round to nearest, symmetric, blocks of 32): matrix multiplications to 4-bit `MatMulNBits`, convolution weights to
24
+ blocked int8 `DequantizeLinear` (opset 21; 4 bits made the vocoder brighter). Activations stay float32.
25
+ 534 MB → 145 MB.
26
+ - `voices/default.bin` (0.26 MB): what onnx-community's `speech_encoder.onnx` (float32) outputs for its
27
+ `default_voice.wav`, computed once, so the browser needs neither the speech encoder (591 MB) nor a reference clip.
28
+ Little-endian float32: N, P, F, then `audio_features` [N, 1024], `audio_tokens` [P] (whole numbers),
29
+ `speaker_embeddings` [192] and `speaker_features` [F, 80]. Made by `tools/export/chatterbox_voice.py` in
30
+ [Quick Inference](https://github.com/mdanyalmalik/quick-inf).
31
+
32
+ The decoder is a modified file: its output is close to, but not identical to, the original's. On eight test
33
+ sentences, Whisper small.en heard it with a 4.5% word error rate, the same as the original float32 graphs.
34
+
35
+ ## Use
36
+
37
+ As the ONNX example in onnx-community's model card, with this file's four arrays in place of the speech encoder's
38
+ outputs (`audio_features` before the text's embeddings in the first language-model step, exaggeration 0.5, greedy
39
+ speech tokens with repetition penalty 1.2 until token 6562) and this decoder: `audio_tokens` and the new tokens with
40
+ the two speaker arrays → 24 kHz audio. Quick Inference speaks one sentence at a time and runs the text through the
41
+ Python package's `punc_norm` first.
42
+
43
+ ## Licence
44
+
45
+ MIT, as Chatterbox (`LICENSE`).
onnx/conditional_decoder.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4bc107a9996b313f5d1a094cdbccd9afaa20e6be596170eac74d8b0d903f1663
3
+ size 145358459
voices/default.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1cb6ec75b9d8e58c4c5d7936c887307fe16c7b76bd91bbeda9e503fc9a9f8191
3
+ size 255732