DevKB/bark
0
1import numpy as np2import gradio as gr3from bark import SAMPLE_RATE, generate_audio, preload_models4from bark.generation import SUPPORTED_LANGS5from share_btn import community_icon_html, loading_icon_html, share_js6 7DEBUG_MODE = False8 9if not DEBUG_MODE:10 _ = preload_models()11 12AVAILABLE_PROMPTS = ["Unconditional", "Announcer"]13PROMPT_LOOKUP = {}14for _, lang in SUPPORTED_LANGS:15 for n in range(10):16 label = f"Speaker {n} ({lang})"17 AVAILABLE_PROMPTS.append(label)18 PROMPT_LOOKUP[label] = f"{lang}_speaker_{n}"19PROMPT_LOOKUP["Unconditional"] = None20PROMPT_LOOKUP["Announcer"] = "announcer"21 22default_text = "Hello, my name is Suno. And, uh — and I like pizza. [laughs]\nBut I also have other interests such as playing tic tac toe."23 24title = "# 🐶 Bark</div>"25 26description = """27<div>28<a style="display:inline-block" href='https://github.com/suno-ai/bark'><img src='https://img.shields.io/github/stars/suno-ai/bark?style=social' /></a>29<a style='display:inline-block' href='https://discord.gg/J2B2vsjKuE'><img src='https://dcbadge.vercel.app/api/server/J2B2vsjKuE?compact=true&style=flat' /></a>30<a style="display:inline-block; margin-left: 1em" href="https://huggingface.co/spaces/suno/bark?duplicate=true"><img src="https://img.shields.io/badge/-Duplicate%20Space%20to%20skip%20the%20queue-blue?labelColor=white&style=flat&logo=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAP5JREFUOE+lk7FqAkEURY+ltunEgFXS2sZGIbXfEPdLlnxJyDdYB62sbbUKpLbVNhyYFzbrrA74YJlh9r079973psed0cvUD4A+4HoCjsA85X0Dfn/RBLBgBDxnQPfAEJgBY+A9gALA4tcbamSzS4xq4FOQAJgCDwV2CPKV8tZAJcAjMMkUe1vX+U+SMhfAJEHasQIWmXNN3abzDwHUrgcRGmYcgKe0bxrblHEB4E/pndMazNpSZGcsZdBlYJcEL9Afo75molJyM2FxmPgmgPqlWNLGfwZGG6UiyEvLzHYDmoPkDDiNm9JR9uboiONcBXrpY1qmgs21x1QwyZcpvxt9NS09PlsPAAAAAElFTkSuQmCC&logoWidth=14" alt="Duplicate Space"></a>31</div>32Bark is a universal text-to-audio model created by [Suno](www.suno.ai), with code publicly available [here](https://github.com/suno-ai/bark). \33Bark can generate highly realistic, multilingual speech as well as other audio - including music, background noise and simple sound effects. \34This demo should be used for research purposes only. Commercial use is strictly prohibited. \35The model output is not censored and the authors do not endorse the opinions in the generated content. \36Use at your own risk.37"""38 39article = """40 41## 🌎 Foreign Language42 43Bark supports various languages out-of-the-box and automatically determines language from input text. \44When prompted with code-switched text, Bark will even attempt to employ the native accent for the respective languages in the same voice.45 46Try the prompt:47 48```49Buenos días Miguel. Tu colega piensa que tu alemán es extremadamente malo. But I suppose your english isn't terrible.50```51 52## 🤭 Non-Speech Sounds53 54Below is a list of some known non-speech sounds, but we are finding more every day. \55Please let us know if you find patterns that work particularly well on Discord!56 57* [laughter]58* [laughs]59* [sighs]60* [music]61* [gasps]62* [clears throat]63* — or ... for hesitations64* ♪ for song lyrics65* capitalization for emphasis of a word66* MAN/WOMAN: for bias towards speaker67 68Try the prompt:69 70```71" [clears throat] Hello, my name is Suno. And, uh — and I like pizza. [laughs] But I also have other interests such as... ♪ singing ♪."72```73 74## 🎶 Music75Bark can generate all types of audio, and, in principle, doesn't see a difference between speech and music. \76Sometimes Bark chooses to generate text as music, but you can help it out by adding music notes around your lyrics.77 78Try the prompt:79 80```81♪ In the jungle, the mighty jungle, the lion barks tonight ♪82```83 84## 🧬 Voice Cloning85 86Bark has the capability to fully clone voices - including tone, pitch, emotion and prosody. \87The model also attempts to preserve music, ambient noise, etc. from input audio. \88However, to mitigate misuse of this technology, we limit the audio history prompts to a limited set of Suno-provided, fully synthetic options to choose from.89 90## 👥 Speaker Prompts91 92You can provide certain speaker prompts such as NARRATOR, MAN, WOMAN, etc. \93Please note that these are not always respected, especially if a conflicting audio history prompt is given.94 95Try the prompt:96 97```98WOMAN: I would like an oatmilk latte please.99MAN: Wow, that's expensive!100```101 102## Details103 104Bark model by [Suno](https://suno.ai/), including official [code](https://github.com/suno-ai/bark) and model weights. \105Gradio demo supported by 🤗 Hugging Face. Bark is licensed under a non-commercial license: CC-BY 4.0 NC, see details on [GitHub](https://github.com/suno-ai/bark).106 107 108"""109 110examples = [111 ["Please surprise me and speak in whatever voice you enjoy. Vielen Dank und Gesundheit!",112 "Unconditional"], # , 0.7, 0.7],113 ["Hello, my name is Suno. And, uh — and I like pizza. [laughs] But I also have other interests such as playing tic tac toe.",114 "Speaker 1 (en)"], # , 0.7, 0.7],115 ["Buenos días Miguel. Tu colega piensa que tu alemán es extremadamente malo. But I suppose your english isn't terrible.",116 "Speaker 0 (es)"], # , 0.7, 0.7],117]118 119 120def gen_tts(text, history_prompt): # , temp_semantic, temp_waveform):121 history_prompt = PROMPT_LOOKUP[history_prompt]122 if DEBUG_MODE:123 audio_arr = np.zeros(SAMPLE_RATE)124 else:125 # , text_temp=temp_semantic, waveform_temp=temp_waveform)126 audio_arr = generate_audio(text, history_prompt=history_prompt)127 audio_arr = (audio_arr * 32767).astype(np.int16)128 return (SAMPLE_RATE, audio_arr)129 130 131css = """132 #share-btn-container {133 display: flex;134 padding-left: 0.5rem !important;135 padding-right: 0.5rem !important;136 background-color: #000000;137 justify-content: center;138 align-items: center;139 border-radius: 9999px !important; 140 width: 13rem;141 margin-top: 10px;142 margin-left: auto;143 flex: unset !important;144 }145 #share-btn {146 all: initial;147 color: #ffffff;148 font-weight: 600;149 cursor: pointer;150 font-family: 'IBM Plex Sans', sans-serif;151 margin-left: 0.5rem !important;152 padding-top: 0.25rem !important;153 padding-bottom: 0.25rem !important;154 right:0;155 }156 #share-btn * {157 all: unset !important;158 }159 #share-btn-container div:nth-child(-n+2){160 width: auto !important;161 min-height: 0px !important;162 }163 #share-btn-container .wrap {164 display: none !important;165 }166"""167with gr.Blocks(css=css) as block:168 gr.Markdown(title)169 gr.Markdown(description)170 with gr.Row():171 with gr.Column():172 input_text = gr.Textbox(173 label="Input Text", lines=2, value=default_text, elem_id="input_text")174 options = gr.Dropdown(175 AVAILABLE_PROMPTS, value="Speaker 1 (en)", label="Acoustic Prompt", elem_id="speaker_option")176 run_button = gr.Button(text="Generate Audio", type="button")177 with gr.Column():178 audio_out = gr.Audio(label="Generated Audio",179 type="numpy", elem_id="audio_out")180 with gr.Row(visible=False) as share_row:181 with gr.Group(elem_id="share-btn-container"):182 community_icon = gr.HTML(community_icon_html)183 loading_icon = gr.HTML(loading_icon_html)184 share_button = gr.Button(185 "Share to community", elem_id="share-btn")186 share_button.click(None, [], [], _js=share_js)187 inputs = [input_text, options]188 outputs = [audio_out]189 gr.Examples(examples=examples, fn=gen_tts, inputs=inputs,190 outputs=outputs, cache_examples=True)191 gr.Markdown(article)192 run_button.click(fn=lambda: gr.update(visible=False), inputs=None, outputs=share_row, queue=False).then(193 fn=gen_tts, inputs=inputs, outputs=outputs, queue=True).then(194 fn=lambda: gr.update(visible=True), inputs=None, outputs=share_row, queue=False)195 196block.queue()197block.launch()198