β‘ Low GPU memory? No problem. CPU offloading. β‘ Somewhat easy install?
(Mamba is a fast version of conda. They should work the same.)
Pip and conda/mamba are two different ways of installing Bark Infinity. If you use Mamba do not install anything. Don't install pytorch, do not install anything with 'CUDA' in the same. You don't need to lookup a YouTube tutorial. Just type the commands. The only thing you need installed is the NVIDIA drivers. If you have a mac, use environment-cpu.yml instead of environment-cpu.yml
There is one exception, on Windows if you don't have the better Windows Terminal installed, that is a nice to have feature https://apps.microsoft.com/store/detail/windows-terminal/9N0DX20HK701
You don't have to but it may dispaly the output from the bark commands better. When you start Anaconda Prompt (miniconda3) you can do it from the new Windows Terminal app, clicking on the down arrow next to the plus, should let you pick Anaconda Prompt (miniconda3)
(There is a requirements-pip.txt file but I have not tested it currently. This is an alternative install.)
-
Go here: https://github.com/conda-forge/miniforge#mambaforge
-
Download a Python 3.10 Miniconda3 installer for your OS. Windows 64-bit, macOS, and Linux probably don't need a guide. a. Install the Mambaforge for your OS, not specifically Windows. OSX for OSX etc. b. Don't install Mambaforge-pypy3. (It probably works fine, it is just not what I tested.) Install the one above that, just plain Mambaforge. Or you can use Conda, Mamba should faster but sometimes Conda may be more compatible.
-
Install the Python 3.10 Miniconda3 exe. Then start the miniforge 'Miniforge Prompt Terminal which is a new program it installed. You will always use this program for Bark.
-
Start 'Miniforge Prompt Be careful not to start the regular windows command line. (Unless you installed the new Terminal and know how to switch.) It should say "Anaconda Prompt (miniconda3)"
You should see also terminal that says "(base)".
- Choose the place to install Bark Infinity directory. You can also just leave it at default. If you make a LOT of audio you think about a place with a lot of space.
When you start "Anaconda Prompt (miniconda3)" you will be in a directory, in Windows, probably something like** "C:\Users\YourName"**. Okay to install there. Just remember where you put it. It will be in /bark. (If you already had bark-infinity installed and want to update instead of reinstalling, skip to the end.)
- Type the next commands exactly. Hit "Y" for yes where you need to:
mamba update -y mamba
mamba install -y git
git clone https://github.com/JonathanFly/bark.git
cd bark
mamba env create -f environment-cuda.yml
mamba activate bark-infinity-oneclick
python -m pip install --upgrade pip
pip install --upgrade setuptools
pip install -r requirements_conda_missing.txt
python bark_perform.py
python bark_webui.py
(If you see a warning that "No GPU being used. Careful, inference might be very slow!" after python bark_perform.py then something may be wrong, if you have GPU. If you didn't see that then the GPU is working.)
To restart later, start Miniforge Prompt. Not Regular Prompt. Make sure you see (base) You will type a command to activate bark-infinity-oneclick and of base, like this:
mamba activate bark-infinity-oneclick
cd bark
python bark_webui.py
git pull
mamba env update -f environment-cuda.yml --prune
python -m pip install --upgrade pip
pip install --upgrade setuptools
pip install -r requirements_conda_missing.txt
!git clone https://github.com/JonathanFly/bark.git
%cd bark
!pip install -r requirements-pip.txt
IF you get a strange error during pip install -r requirements_conda_missing.txt and are on a Windows computer type this:
pip install fairseq@https://github.com/Sharrnah/fairseq/releases/download/v0.12.4/fairseq-0.12.4-cp310-cp310-win_amd64.whl
If you are on a mac, you may need to just type mamba install fairseq (the windows conda version is too out of date)
I have so much good Bark I need to post at twitter.com/jonathanfly
Bark Infinity started as a humble π» command line wrapper, a CLI π¬. Built from simple keyword commands, it was a proof of concept π§ͺ, a glimmer of potential π‘.
Bark Infinity evolved π§¬, expanding across dimensions π. Infinite Length π΅π, Infinite Voices ππ, and a true high point in human history: π Infinite Awkwardness πΊ. But for some people, the time-tested command line interface was not a good fit. Many couldn't even try Bark π, struggling with CUDA gods π© and being left with cryptic error messages π§ and a chaotic computer πΎ. Many people felt veryβ¦ UN INFINITE.
π Bark Infinity πΎ was born in the command line, and Bark Infinity grew within the command line. We live in the era where old fashioned command line applications are wrapped in β¨fancy Gradio Uisπ and π±οΈOne Click Installers. We all must adapt to a changing world, right? Or do we?
Discover cool new voices and reuse them. Performers, musicians, sound effects, two party dialog scenes. Save and share them. Every audio clip saves a speaker.npz file with the voice. To reuse a voice, move the generated speaker.npz file (named the same as the .wav file) to the "prompts" directory inside "bark" where all the other .npz files are.
π With random celebrity appearances!
(I accidentally left a bunch of voices in the repo, some of them are pretty good. Use --history_prompt 'en_fiery' for the same voice as the audio sample right after this sentence.)
whoami.mp4
Any length prompt and audio clips. Sometimes the final result is seamless, sometimes it's stable (but usually not both!).
π΅ Now with Slowly Morphing Rick Rolls! Can you even spot the seams in the most earnest Rick Rolls you've ever heard in your life?
but_are_we_strangers_to_love_really.mp4
Confused Travolta GIF
Can your text-to-speech model stammer and stall like a student answering a question about a book they didn't read? Bark can. That's the human touch. The semantic touch. You can almost feel the awkward silence through the screen.
Are you tired of telling your TTS model what to say? Why not take a break and let your TTS model do the work for you. With enough patience and Confused Travolta Mode, Bark can finish your jokes for you.
almost_a_real_joke.mp4
Truly we live in the future. It might take 50 tries to get a joke and it's probably an accident, but all 49 failures are also very amusing so it's a win/win. (That's right, I set a single function flag to False in a Bark and raved about the amazing new feature. Everything here is small potatoes really.)
reaching_for_the_words.mp4
BARK INFINITY is possible because Bark is such an amazingly simple and powerful model that even I could poke around easily.
For music, I recommend using the --split_by_lines and making sure you use a multiline string as input. You'll generally get better results if you manually split your text, which I neglected to provide an easy way to do because I stayed too late listening to 100 different Bark versions of a scene an Andor and failed Why was 6 afraid of 7 jokes.
Type --help or use the GUI
Usage: bark_perform.py [-h] [--text_prompt TEXT_PROMPT] [--list_speakers LIST_SPEAKERS] [--dry_run DRY_RUN] [--text_splits_only TEXT_SPLITS_ONLY] [--history_prompt HISTORY_PROMPT]
[--prompt_file PROMPT_FILE] [--split_input_into_separate_prompts_by {word,line,sentence,char,string,random,regex}]
[--split_input_into_separate_prompts_by_value SPLIT_INPUT_INTO_SEPARATE_PROMPTS_BY_VALUE] [--always_save_speaker ALWAYS_SAVE_SPEAKER]
[--output_iterations OUTPUT_ITERATIONS] [--output_filename OUTPUT_FILENAME] [--output_dir OUTPUT_DIR] [--hoarder_mode HOARDER_MODE] [--extra_stats EXTRA_STATS]
[--show_generation_times SHOW_GENERATION_TIMES] [--output_format {wav,mp3,ogg,flac,mp4}] [--text_use_gpu TEXT_USE_GPU] [--text_use_small TEXT_USE_SMALL]
[--coarse_use_gpu COARSE_USE_GPU] [--coarse_use_small COARSE_USE_SMALL] [--fine_use_gpu FINE_USE_GPU] [--fine_use_small FINE_USE_SMALL] [--codec_use_gpu CODEC_USE_GPU]
[--force_reload FORCE_RELOAD] [--GLOBAL_ENABLE_MPS GLOBAL_ENABLE_MPS] [--USE_SMALL_MODELS USE_SMALL_MODELS] [--OFFLOAD_CPU OFFLOAD_CPU] [--text_temp TEXT_TEMP]
[--waveform_temp WAVEFORM_TEMP] [--confused_travolta_mode CONFUSED_TRAVOLTA_MODE] [--silent SILENT] [--seed SEED] [--stable_mode_interval STABLE_MODE_INTERVAL]
[--single_starting_seed SINGLE_STARTING_SEED] [--split_character_goal_length SPLIT_CHARACTER_GOAL_LENGTH] [--split_character_max_length SPLIT_CHARACTER_MAX_LENGTH]
[--split_character_jitter SPLIT_CHARACTER_JITTER] [--add_silence_between_segments ADD_SILENCE_BETWEEN_SEGMENTS]
[--process_text_by_each {word,line,sentence,char,string,random,regex}] [--group_text_by_counting {word,line,sentence,char,string,random,regex}]
[--in_groups_of_size IN_GROUPS_OF_SIZE] [--split_type_string SPLIT_TYPE_STRING] [--prompt_text_prefix PROMPT_TEXT_PREFIX]
[--extra_confused_travolta_mode EXTRA_CONFUSED_TRAVOLTA_MODE] [--seperate_prompts SEPERATE_PROMPTS] [--semantic_history_only SEMANTIC_HISTORY_ONLY]
[--absolute_semantic_history_only ABSOLUTE_SEMANTIC_HISTORY_ONLY] [--absolute_semantic_history_only_every_x ABSOLUTE_SEMANTIC_HISTORY_ONLY_EVERY_X]
[--semantic_history_starting_weight SEMANTIC_HISTORY_STARTING_WEIGHT] [--semantic_history_future_weight SEMANTIC_HISTORY_FUTURE_WEIGHT]
[--semantic_prev_segment_weight SEMANTIC_PREV_SEGMENT_WEIGHT] [--coarse_history_starting_weight COARSE_HISTORY_STARTING_WEIGHT]
[--coarse_history_future_weight COARSE_HISTORY_FUTURE_WEIGHT] [--coarse_prev_segment_weight COARSE_PREV_SEGMENT_WEIGHT]
[--fine_history_starting_weight FINE_HISTORY_STARTING_WEIGHT] [--fine_history_future_weight FINE_HISTORY_FUTURE_WEIGHT]
[--fine_prev_segment_weight FINE_PREV_SEGMENT_WEIGHT] [--custom_audio_processing_function CUSTOM_AUDIO_PROCESSING_FUNCTION] [--use_smaller_models USE_SMALLER_MODELS]
[--bark_cloning_large_model BARK_CLONING_LARGE_MODEL] [--semantic_temp SEMANTIC_TEMP] [--semantic_top_k SEMANTIC_TOP_K] [--semantic_top_p SEMANTIC_TOP_P]
[--semantic_min_eos_p SEMANTIC_MIN_EOS_P] [--semantic_max_gen_duration_s SEMANTIC_MAX_GEN_DURATION_S] [--semantic_allow_early_stop SEMANTIC_ALLOW_EARLY_STOP]
[--semantic_use_kv_caching SEMANTIC_USE_KV_CACHING] [--semantic_seed SEMANTIC_SEED] [--semantic_history_oversize_limit SEMANTIC_HISTORY_OVERSIZE_LIMIT]
[--coarse_temp COARSE_TEMP] [--coarse_top_k COARSE_TOP_K] [--coarse_top_p COARSE_TOP_P] [--coarse_max_coarse_history COARSE_MAX_COARSE_HISTORY]
[--coarse_sliding_window_len COARSE_SLIDING_WINDOW_LEN] [--coarse_kv_caching COARSE_KV_CACHING] [--coarse_seed COARSE_SEED]
[--x_coarse_history_alignment_hack X_COARSE_HISTORY_ALIGNMENT_HACK] [--fine_temp FINE_TEMP] [--fine_seed FINE_SEED] [--render_npz_samples RENDER_NPZ_SAMPLES]
This is the first prompt.
Lots of text here maybe.
As long as you want.
AAAAA
This is the second prompt.
AAAAA
This is the third prompt.
AAAAA
This is the fourth prompt.python bark_perform.py --prompt_file myprompts.txt --split_input_into_separate_prompts_by string --split_input_into_separate_prompts_by_value AAAAA --output_dir myprompts_samples
Examples β’ Suno Studio Waitlist β’ Updates β’ How to Use β’ Installation β’ FAQ
Bark is a transformer-based text-to-audio model created by Suno. Bark can generate highly realistic, multilingual speech as well as other audio - including music, background noise and simple sound effects. The model can also produce nonverbal communications like laughing, sighing and crying. To support the research community, we are providing access to pretrained model checkpoints, which are ready for inference and available for commercial use.
Bark was developed for research purposes. It is not a conventional text-to-speech model but instead a fully generative text-to-audio model, which can deviate in unexpected ways from provided prompts. Suno does not take responsibility for any output generated. Use at your own risk, and please act responsibly.
2023.05.01
-
Β©οΈ Bark is now licensed under the MIT License, meaning it's now available for commercial use!
-
β‘ 2x speed-up on GPU. 10x speed-up on CPU. We also added an option for a smaller version of Bark, which offers additional speed-up with the trade-off of slightly lower quality.
-
π Long-form generation, voice consistency enhancements and other examples are now documented in a new notebooks section.
-
π₯ We created a voice prompt library. We hope this resource helps you find useful prompts for your use cases! You can also join us on Discord, where the community actively shares useful prompts in the #audio-prompts channel.
-
π¬ Growing community support and access to new features here:
-
πΎ You can now use Bark with GPUs that have low VRAM (<4GB).
2023.04.20
- πΆ Bark release!
from bark import SAMPLE_RATE, generate_audio, preload_models
from scipy.io.wavfile import write as write_wav
from IPython.display import Audio
# download and load all models
preload_models()
# generate audio from text
text_prompt = """
Hello, my name is Suno. And, uh β and I like pizza. [laughs]
But I also have other interests such as playing tic tac toe.
"""
audio_array = generate_audio(text_prompt)
# save audio to disk
write_wav("bark_generation.wav", SAMPLE_RATE, audio_array)
# play text in notebook
Audio(audio_array, rate=SAMPLE_RATE)pizza.webm
Bark supports various languages out-of-the-box and automatically determines language from input text. When prompted with code-switched text, Bark will attempt to employ the native accent for the respective languages. English quality is best for the time being, and we expect other languages to further improve with scaling.
text_prompt = """
μΆμμ λ΄κ° κ°μ₯ μ’μνλ λͺ
μ μ΄λ€. λλ λ©°μΉ λμ ν΄μμ μ·¨νκ³ μΉκ΅¬ λ° κ°μ‘±κ³Ό μκ°μ λ³΄λΌ μ μμ΅λλ€.
"""
audio_array = generate_audio(text_prompt)suno_korean.webm
Note: since Bark recognizes languages automatically from input text, it is possible to use for example a german history prompt with english text. This usually leads to english audio with a german accent.
Bark can generate all types of audio, and, in principle, doesn't see a difference between speech and music. Sometimes Bark chooses to generate text as music, but you can help it out by adding music notes around your lyrics.
text_prompt = """
βͺ In the jungle, the mighty jungle, the lion barks tonight βͺ
"""
audio_array = generate_audio(text_prompt)lion.webm
Bark supports 100+ speaker presets across supported languages. You can browse the library of speaker presets here, or in the code. The community also often shares presets in Discord.
Bark tries to match the tone, pitch, emotion and prosody of a given preset, but does not currently support custom voice cloning. The model also attempts to preserve music, ambient noise, etc.
text_prompt = """
I have a silky smooth voice, and today I will tell you about
the exercise regimen of the common sloth.
"""
audio_array = generate_audio(text_prompt, history_prompt="v2/en_speaker_1")sloth.webm
By default, generate_audio works well with around 13 seconds of spoken text. For an example of how to do long-form generation, see this example notebook.
Click to toggle example long-form generations (from the example notebook)
dialog.webm
longform_advanced.webm
longform_basic.webm
pip install git+https://github.com/suno-ai/bark.git
or
git clone https://github.com/suno-ai/bark
cd bark && pip install .
Note: Do NOT use 'pip install bark'. It installs a different package, which is not managed by Suno.
Bark has been tested and works on both CPU and GPU (pytorch 2.0+, CUDA 11.7 and CUDA 12.0).
On enterprise GPUs and PyTorch nightly, Bark can generate audio in roughly real-time. On older GPUs, default colab, or CPU, inference time might be significantly slower. For older GPUs or CPU you might want to consider using smaller models. Details can be found in out tutorial sections here.
The full version of Bark requires around 12GB of VRAM to hold everything on GPU at the same time.
To use a smaller version of the models, which should fit into 8GB VRAM, set the environment flag SUNO_USE_SMALL_MODELS=True.
If you don't have hardware available or if you want to play with bigger versions of our models, you can also sign up for early access to our model playground here.
Bark is fully generative tex-to-audio model devolved for research and demo purposes. It follows a GPT style architecture similar to AudioLM and Vall-E and a quantized Audio representation from EnCodec. It is not a conventional TTS model, but instead a fully generative text-to-audio model capable of deviating in unexpected ways from any given script. Different to previous approaches, the input text prompt is converted directly to audio without the intermediate use of phonemes. It can therefore generalize to arbitrary instructions beyond speech such as music lyrics, sound effects or other non-speech sounds.
Below is a list of some known non-speech sounds, but we are finding more every day. Please let us know if you find patterns that work particularly well on Discord!
[laughter][laughs][sighs][music][gasps][clears throat]βor...for hesitationsβͺfor song lyrics- CAPITALIZATION for emphasis of a word
[MAN]and[WOMAN]to bias Bark toward male and female speakers, respectively
| Language | Status |
|---|---|
| English (en) | β |
| German (de) | β |
| Spanish (es) | β |
| French (fr) | β |
| Hindi (hi) | β |
| Italian (it) | β |
| Japanese (ja) | β |
| Korean (ko) | β |
| Polish (pl) | β |
| Portuguese (pt) | β |
| Russian (ru) | β |
| Turkish (tr) | β |
| Chinese, simplified (zh) | β |
Requests for future language support here or in the #forums channel on Discord.
- nanoGPT for a dead-simple and blazing fast implementation of GPT-style models
- EnCodec for a state-of-the-art implementation of a fantastic audio codec
- AudioLM for related training and inference code
- Vall-E, AudioLM and many other ground-breaking papers that enabled the development of Bark
Bark is licensed under the MIT License.
Please contact us at bark@suno.ai to request access to a larger version of the model.
Weβre developing a playground for our models, including Bark.
If you are interested, you can sign up for early access here.
- Bark uses Hugging Face to download and store models. You can see find more info here.
- Bark is a GPT-style model. As such, it may take some creative liberties in its generations, resulting in higher-variance model outputs than traditional text-to-speech approaches.
- Bark supports 100+ speaker presets across supported languages. You can browse the library of speaker presets here. The community also shares presets in Discord. Bark also supports generating unique random voices that fit the input text. Bark does not currently support custom voice cloning.
- Bark is a GPT-style model, and its architecture/context window is optimized to output generations with roughly this length.
- The full version of Bark requires around 12Gb of memory to hold everything on GPU at the same time. However, even smaller cards down to ~2Gb work with some additional settings. Simply add the following code snippet before your generation:
import os
os.environ["SUNO_OFFLOAD_CPU"] = True
os.environ["SUNO_USE_SMALL_MODELS"] = True- Bark generates audio from scratch. It is not meant to create only high-fidelity, studio-quality speech. Rather, outputs could be anything from perfect speech to multiple people arguing at a baseball game recorded with bad microphones.