返回 F5-TTS
SHARED.md
根目录 / src / f5_tts / infer / SHARED.md
1 <!-- omit in toc -->
2 # Shared Model Cards
3
4 <!-- omit in toc -->
5 ### **Prerequisites of using**
6 - This document is serving as a quick lookup table for the community training/finetuning result, with various language support.
7 - The models in this repository are open source and are based on voluntary contributions from contributors.
8 - The use of models must be conditioned on respect for the respective creators. The convenience brought comes from their efforts.
9
10 <!-- omit in toc -->
11 ### **Welcome to share here**
12 - Have a pretrained/finetuned result: model checkpoint (pruned best to facilitate inference, i.e. leave only `ema_model_state_dict`) and corresponding vocab file (for tokenization).
13 - Host a public [huggingface model repository](https://huggingface.co/new) and upload the model related files.
14 - Make a pull request adding a model card to the current page, i.e. `src\f5_tts\infer\SHARED.md`.
15
16 <!-- omit in toc -->
17 ### Supported Languages
18 - [Multilingual](#multilingual)
19 - [F5-TTS v1 v0 Base @ zh \& en @ F5-TTS](#f5-tts-v1-v0-base--zh--en--f5-tts)
20 - [Arabic](#arabic)
21 - [F5-TTS Small @ ar & en @ SILMA AI](#f5-tts-small--ar--en--silma-ai)
22 - [English](#english)
23 - [Finnish](#finnish)
24 - [F5-TTS Base @ fi @ AsmoKoskinen](#f5-tts-base--fi--asmokoskinen)
25 - [French](#french)
26 - [F5-TTS Base @ fr @ RASPIAUDIO](#f5-tts-base--fr--raspiaudio)
27 - [German](#german)
28 - [F5-TTS Base @ de @ hvoss-techfak](#f5-tts-base--de--hvoss-techfak)
29 - [Hindi](#hindi)
30 - [F5-TTS Small @ hi @ SPRINGLab](#f5-tts-small--hi--springlab)
31 - [Italian](#italian)
32 - [F5-TTS Base @ it @ alien79](#f5-tts-base--it--alien79)
33 - [Japanese](#japanese)
34 - [F5-TTS Base @ ja @ Jmica](#f5-tts-base--ja--jmica)
35 - [Latvian](#latvian)
36 - [F5-TTS Base @ lv @ RaivisDejus](#f5-tts-base--lv--raivisdejus)
37 - [Mandarin](#mandarin)
38 - [Russian](#russian)
39 - [F5-TTS Base @ ru @ HotDro4illa](#f5-tts-base--ru--hotdro4illa)
40 - [Spanish](#spanish)
41 - [F5-TTS Base @ es @ jpgallegoar](#f5-tts-base--es--jpgallegoar)
42
43
44 ## Multilingual
45
46 #### F5-TTS v1 v0 Base @ zh & en @ F5-TTS
47 |Model|🤗Hugging Face|Data (Hours)|Model License|
48 |:---:|:------------:|:-----------:|:-------------:|
49 |F5-TTS v1 Base|[ckpt & vocab](https://huggingface.co/SWivid/F5-TTS/tree/main/F5TTS_v1_Base)|[Emilia 95K zh&en](https://huggingface.co/datasets/amphion/Emilia-Dataset/tree/fc71e07)|cc-by-nc-4.0|
50
51 ```bash
52 Model: hf://SWivid/F5-TTS/F5TTS_v1_Base/model_1250000.safetensors
53 # A Variant Model: hf://SWivid/F5-TTS/F5TTS_v1_Base_no_zero_init/model_1250000.safetensors
54 Vocab: hf://SWivid/F5-TTS/F5TTS_v1_Base/vocab.txt
55 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "conv_layers": 4}
56 ```
57
58 |Model|🤗Hugging Face|Data (Hours)|Model License|
59 |:---:|:------------:|:-----------:|:-------------:|
60 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/SWivid/F5-TTS/tree/main/F5TTS_Base)|[Emilia 95K zh&en](https://huggingface.co/datasets/amphion/Emilia-Dataset/tree/fc71e07)|cc-by-nc-4.0|
61
62 ```bash
63 Model: hf://SWivid/F5-TTS/F5TTS_Base/model_1200000.safetensors
64 Vocab: hf://SWivid/F5-TTS/F5TTS_Base/vocab.txt
65 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
66 ```
67
68 *Other infos, e.g. Author info, Github repo, Link to some sampled results, Usage instruction, Tutorial (Blog, Video, etc.) ...*
69
70
71 ## Arabic
72
73 #### F5-TTS Small @ ar & en @ SILMA AI
74 |Model|🤗Hugging Face|Data (Hours)|Model License|
75 |:---:|:------------:|:-----------:|:-------------:|
76 |F5-TTS Small|[ckpt & vocab](https://huggingface.co/silma-ai/silma-tts)| Tens of thousands EN/AR |Apache-2.0|
77
78 - Pretrained by [SILMA.AI](https://silma.ai)
79 - [GitHub repo](https://github.com/SILMA-AI/silma-tts), Inference code
80
81
82 ## English
83
84
85 ## Finnish
86
87 #### F5-TTS Base @ fi @ AsmoKoskinen
88 |Model|🤗Hugging Face|Data|Model License|
89 |:---:|:------------:|:-----------:|:-------------:|
90 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/AsmoKoskinen/F5-TTS_Finnish_Model)|[Common Voice](https://huggingface.co/datasets/mozilla-foundation/common_voice_17_0), [Vox Populi](https://huggingface.co/datasets/facebook/voxpopuli)|cc-by-nc-4.0|
91
92 ```bash
93 Model: hf://AsmoKoskinen/F5-TTS_Finnish_Model/model_common_voice_fi_vox_populi_fi_20241206.safetensors
94 Vocab: hf://AsmoKoskinen/F5-TTS_Finnish_Model/vocab.txt
95 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
96 ```
97
98
99 ## French
100
101 #### F5-TTS Base @ fr @ RASPIAUDIO
102 |Model|🤗Hugging Face|Data (Hours)|Model License|
103 |:---:|:------------:|:-----------:|:-------------:|
104 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/RASPIAUDIO/F5-French-MixedSpeakers-reduced)|[LibriVox](https://librivox.org/)|cc-by-nc-4.0|
105
106 ```bash
107 Model: hf://RASPIAUDIO/F5-French-MixedSpeakers-reduced/model_last_reduced.pt
108 Vocab: hf://RASPIAUDIO/F5-French-MixedSpeakers-reduced/vocab.txt
109 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
110 ```
111
112 - [Online Inference with Hugging Face Space](https://huggingface.co/spaces/RASPIAUDIO/f5-tts_french).
113 - [Tutorial video to train a new language model](https://www.youtube.com/watch?v=UO4usaOojys).
114 - [Discussion about this training can be found here](https://github.com/SWivid/F5-TTS/issues/434).
115
116
117 ## German
118
119 #### F5-TTS Base @ de @ hvoss-techfak
120 |Model|🤗Hugging Face|Data (Hours)|Model License|
121 |:---:|:------------:|:-----------:|:-------------:|
122 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/hvoss-techfak/F5-TTS-German)|[Mozilla Common Voice 19.0](https://commonvoice.mozilla.org/en/datasets) & 800 hours Crowdsourced |cc-by-nc-4.0|
123
124 ```bash
125 Model: hf://hvoss-techfak/F5-TTS-German/model_f5tts_german.pt
126 Vocab: hf://hvoss-techfak/F5-TTS-German/vocab.txt
127 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
128 ```
129
130 - Finetuned by [@hvoss-techfak](https://github.com/hvoss-techfak)
131
132
133 ## Hindi
134
135 #### F5-TTS Small @ hi @ SPRINGLab
136 |Model|🤗Hugging Face|Data (Hours)|Model License|
137 |:---:|:------------:|:-----------:|:-------------:|
138 |F5-TTS Small|[ckpt & vocab](https://huggingface.co/SPRINGLab/F5-Hindi-24KHz)|[IndicTTS Hi](https://huggingface.co/datasets/SPRINGLab/IndicTTS-Hindi) & [IndicVoices-R Hi](https://huggingface.co/datasets/SPRINGLab/IndicVoices-R_Hindi) |cc-by-4.0|
139
140 ```bash
141 Model: hf://SPRINGLab/F5-Hindi-24KHz/model_2500000.safetensors
142 Vocab: hf://SPRINGLab/F5-Hindi-24KHz/vocab.txt
143 Config: {"dim": 768, "depth": 18, "heads": 12, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
144 ```
145
146 - Authors: SPRING Lab, Indian Institute of Technology, Madras
147 - Website: https://asr.iitm.ac.in/
148
149
150 ## Italian
151
152 #### F5-TTS Base @ it @ alien79
153 |Model|🤗Hugging Face|Data|Model License|
154 |:---:|:------------:|:-----------:|:-------------:|
155 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/alien79/F5-TTS-italian)|[ylacombe/cml-tts](https://huggingface.co/datasets/ylacombe/cml-tts) |cc-by-nc-4.0|
156
157 ```bash
158 Model: hf://alien79/F5-TTS-italian/model_159600.safetensors
159 Vocab: hf://alien79/F5-TTS-italian/vocab.txt
160 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
161 ```
162
163 - Trained by [Mithril Man](https://github.com/MithrilMan)
164 - Model details on [hf project home](https://huggingface.co/alien79/F5-TTS-italian)
165 - Open to collaborations to further improve the model
166
167
168 ## Japanese
169
170 #### F5-TTS Base @ ja @ Jmica
171 |Model|🤗Hugging Face|Data (Hours)|Model License|
172 |:---:|:------------:|:-----------:|:-------------:|
173 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/Jmica/F5TTS/tree/main/JA_21999120)|[Emilia 1.7k JA](https://huggingface.co/datasets/amphion/Emilia-Dataset/tree/fc71e07) & [Galgame Dataset 5.4k](https://huggingface.co/datasets/OOPPEENN/Galgame_Dataset)|cc-by-nc-4.0|
174
175 ```bash
176 Model: hf://Jmica/F5TTS/JA_21999120/model_21999120.pt
177 Vocab: hf://Jmica/F5TTS/JA_21999120/vocab_japanese.txt
178 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
179 ```
180
181
182 ## Latvian
183
184 #### F5-TTS Base @ lv @ RaivisDejus
185 |Model|🤗Hugging Face|Data (Hours)|Model License|
186 |:---:|:------------:|:-----------:|:-------------:|
187 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/RaivisDejus/F5-TTS-Latvian)|[Common voice](https://datacollective.mozillafoundation.org/datasets/cmj8u3pec00flnxxbntvfb4as)|cc0-1.0|
188
189 ```bash
190 Model: hf://RaivisDejus/F5-TTS-Latvian/model.safetensors
191 Vocab: hf://RaivisDejus/F5-TTS-Latvian/vocab.txt
192 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
193 ```
194
195
196 ## Mandarin
197
198
199 ## Russian
200
201 #### F5-TTS Base @ ru @ HotDro4illa
202 |Model|🤗Hugging Face|Data (Hours)|Model License|
203 |:---:|:------------:|:-----------:|:-------------:|
204 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/hotstone228/F5-TTS-Russian)|[Common voice](https://huggingface.co/datasets/mozilla-foundation/common_voice_17_0)|cc-by-nc-4.0|
205
206 ```bash
207 Model: hf://hotstone228/F5-TTS-Russian/model_last.safetensors
208 Vocab: hf://hotstone228/F5-TTS-Russian/vocab.txt
209 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
210 ```
211 - Finetuned by [HotDro4illa](https://github.com/HotDro4illa)
212 - Any improvements are welcome
213
214
215 ## Spanish
216
217 #### F5-TTS Base @ es @ jpgallegoar
218 |Model|🤗Hugging Face|Data (Hours)|Model License|
219 |:---:|:------------:|:-----------:|:-------------:|
220 |F5-TTS Base|[ckpt & vocab](https://huggingface.co/jpgallegoar/F5-Spanish)|[Voxpopuli](https://huggingface.co/datasets/facebook/voxpopuli) & Crowdsourced & TEDx, 218 hours|cc0-1.0|
221
222 - @jpgallegoar [GitHub repo](https://github.com/jpgallegoar/Spanish-F5), Jupyter Notebook and Gradio usage for Spanish model.
223
223 lines MARKDOWN