saywoof

commit d6783e28875386e360c76dc2f676c242951a6db6

Author: Connor Etherington <[email protected]>

Date:   Tue Aug 1 22:26:15 2023 +0200

    Auto-Commit Update 01.08.2023 - 22:26:15

---

 recipe/meta.yaml                      |  29 +++++

 saywoof.egg-info/PKG-INFO             |   6 ++

 saywoof.egg-info/SOURCES.txt          |   9 ++

 saywoof.egg-info/dependency_links.txt |   1 +

 saywoof.egg-info/entry_points.txt     |   2 +

 saywoof.egg-info/requires.txt         |   7 ++

 saywoof.egg-info/top_level.txt        |   1 +

 saywoof/__init__.py                   |  18 ++++

 saywoof/main.py                       | 195 ++++++++++++++++++++++++++++++++++

 setup.py                              |  24 +++++

 10 files changed, 292 insertions(+)

diff --git a/recipe/meta.yaml b/recipe/meta.yaml

new file mode 100644

index 0000000..7e56c72

--- /dev/null

+++ b/recipe/meta.yaml

@@ -0,0 +1,29 @@

+package:

+  name: saywoof

+  version: 0.1.0

+

+source:

+  path: ..

+

+build:

+  number: 0

+  script: '{{ PYTHON }} -m pip install . --no-deps -vv'

+

+requirements:

+  host:

+    - python

+    - pip

+  run:

+    - python

+    - "numpy<=1.24"

+    - pandas

+

+about:

+  home: https://gitlab.com/a4to/saywoof

+  license: MIT

+  license_file: LICENSE

+  summary: 'An easy to use wrapper for the BARK text-to-speech engine'

+

+extra:

+  recipe-maintainers:

+    - concise

diff --git a/saywoof.egg-info/PKG-INFO b/saywoof.egg-info/PKG-INFO

new file mode 100644

index 0000000..2067cf6

--- /dev/null

+++ b/saywoof.egg-info/PKG-INFO

@@ -0,0 +1,6 @@

+Metadata-Version: 2.1

+Name: saywoof

+Version: 0.1.0

+Summary: An easy to use wrapper for the BARK text-to-speech engine

+Author: Connor Etherington

+Author-email: [email protected]

diff --git a/saywoof.egg-info/SOURCES.txt b/saywoof.egg-info/SOURCES.txt

new file mode 100644

index 0000000..1eac9c0

--- /dev/null

+++ b/saywoof.egg-info/SOURCES.txt

@@ -0,0 +1,9 @@

+setup.py

+saywoof/__init__.py

+saywoof/main.py

+saywoof.egg-info/PKG-INFO

+saywoof.egg-info/SOURCES.txt

+saywoof.egg-info/dependency_links.txt

+saywoof.egg-info/entry_points.txt

+saywoof.egg-info/requires.txt

+saywoof.egg-info/top_level.txt

diff --git a/saywoof.egg-info/dependency_links.txt b/saywoof.egg-info/dependency_links.txt

new file mode 100644

index 0000000..8b13789

--- /dev/null

+++ b/saywoof.egg-info/dependency_links.txt

@@ -0,0 +1 @@

+

diff --git a/saywoof.egg-info/entry_points.txt b/saywoof.egg-info/entry_points.txt

new file mode 100644

index 0000000..7b5f261

--- /dev/null

+++ b/saywoof.egg-info/entry_points.txt

@@ -0,0 +1,2 @@

+[console_scripts]

+saywoof = saywoof.main:main

diff --git a/saywoof.egg-info/requires.txt b/saywoof.egg-info/requires.txt

new file mode 100644

index 0000000..29024b4

--- /dev/null

+++ b/saywoof.egg-info/requires.txt

@@ -0,0 +1,7 @@

+pyaudio

+numpy

+soundfile

+transformers

+saywoof

+colr

+argparse

diff --git a/saywoof.egg-info/top_level.txt b/saywoof.egg-info/top_level.txt

new file mode 100644

index 0000000..8bcccc1

--- /dev/null

+++ b/saywoof.egg-info/top_level.txt

@@ -0,0 +1 @@

+saywoof

diff --git a/saywoof/__init__.py b/saywoof/__init__.py

new file mode 100644

index 0000000..26d7b22

--- /dev/null

+++ b/saywoof/__init__.py

@@ -0,0 +1,18 @@

+import argparse

+import sys

+

+parser = argparse.ArgumentParser(description='A wrapper for the BARK text-to-speech engine')

+parser.add_argument("-l", "--local", help="Load local model", action="store_true")

+parser.add_argument("-L", "--large", help="Load large model (Uses more vram)", action="store_true")

+parser.add_argument("-s", "--save", help="Save model", action="store_true")

+parser.add_argument("-t", "--text", help="Text to convert to speech")

+parser.add_argument("-f", "--file", help="File to convert to speech")

+parser.add_argument("-o", "--output", help="Output file name (default: output.wav)")

+parser.add_argument("-r", "--rate", help="Sample rate of output file (default: 22050)", type=int)

+parser.add_argument("-p", "--persistent", help="Persistent mode (Keep running after conversions to save time)", action="store_true")

+

+args = parser.parse_args()

+

+if len(sys.argv) == 1:

+    parser.print_help()

+    sys.exit(1)

diff --git a/saywoof/main.py b/saywoof/main.py

new file mode 100644

index 0000000..4b7cae7

--- /dev/null

+++ b/saywoof/main.py

@@ -0,0 +1,195 @@

+import time

+import pyaudio

+import numpy as np

+import soundfile as sf

+from transformers import AutoProcessor, AutoModel, AutoTokenizer

+from saywoof import args

+from colr import color

+import argparse

+import os

+import sys

+

+save_path = None, None

+

+blue_1 = '#67D0A8'

+red_1 = '#FF6B6B'

+

+def pr(text: str, hex_color: str = None):

+

+    if not hex_color:

+        hex_color = blue_1

+

+    print(color(text, fore=hex_color))

+

+

+def load_model():

+    save_path = "./Models/bark-small"

+    os.environ["SUNO_OFFLOAD_CPU"] = "True"

+    os.environ["SUNO_USE_SMALL_MODELS"] = "True"

+

+    pr("\nLoading Processor...\n")

+    processor = AutoProcessor.from_pretrained("suno/bark-small")

+

+    pr("\nLoading Model...\n")

+    model = AutoModel.from_pretrained("suno/bark-small")

+

+    pr("\nLoading Tokenizer...\n")

+    tokenizer = AutoTokenizer.from_pretrained("suno/bark-small")

+

+    return processor, model, tokenizer, save_path

+

+

+def load_large():

+    save_path = "./Models/bark"

+

+    pr("\nLoading Processor...\n")

+    processor = AutoProcessor.from_pretrained("suno/bark")

+

+    pr("\nLoading Model...\n")

+    model = AutoModel.from_pretrained("suno/bark")

+

+    pr("\nLoading Tokenizer...\n")

+    tokenizer = AutoTokenizer.from_pretrained("suno/bark")

+

+    return processor, model, tokenizer, save_path

+

+

+def load_local():

+    os.environ["SUNO_OFFLOAD_CPU"] = "True"

+    os.environ["SUNO_USE_SMALL_MODELS"] = "True"

+

+    if args.large:

+        if not os.path.exists("./Models/bark"):

+            pr("\nBark model not found, please download it first (run without '-l' flag)\n", red_1)

+            sys.exit(1)

+        else:

+            pr("\nLoading Processor...\n")

+            processor = AutoProcessor.from_pretrained("./Models/bark")

+

+            pr("\nLoading Model...\n")

+            model = AutoModel.from_pretrained("./Models/bark")

+

+            pr("\nLoading Tokenizer...\n")

+            tokenizer = AutoTokenizer.from_pretrained("./Models/bark")

+

+    else:

+        if not os.path.exists("./Models/bark-small"):

+            pr("\nBark-small model not found, please download it first (run without '-l' flag)\n", red_1)

+            sys.exit(1)

+        else:

+            pr("\nLoading Processor...\n")

+            processor = AutoProcessor.from_pretrained("./Models/bark-small")

+

+            pr("\nLoading Model...\n")

+            model = AutoModel.from_pretrained("./Models/bark-small")

+

+            pr("\nLoading Tokenizer...\n")

+            tokenizer = AutoTokenizer.from_pretrained("./Models/bark-small")

+

+    return processor, model, tokenizer

+

+

+def save_model(processor, model, tokenizer, save_path):

+

+    if not os.path.exists(save_path):

+        os.makedirs(save_path)

+

+    pr("\nSaving Processor...\n")

+    processor.save_pretrained(save_path)

+

+    pr("\nSaving Model...\n")

+    model.save_pretrained(save_path)

+

+    pr("\nSaving Tokenizer...\n")

+    tokenizer.save_pretrained(save_path)

+

+

+def text_to_speech(processor, model, tokenizer, text):

+

+    voice_preset = "v2/en_speaker_9"

+

+    inputs = processor(

+        text=[text],

+        return_tensors="pt",

+        voice_preset=voice_preset

+    )

+

+    # speech_values = model.generate(**inputs, do_sample=True)

+    audio_array = model.generate(**inputs)

+    audio_array = audio_array.cpu().numpy().squeeze()

+

+    return audio_array

+

+

+def write(filename, sample_rate, audio_array):

+    sf.write(filename, audio_array, sample_rate)

+

+

+def main():

+

+    if args.rate:

+        rate = int(args.rate)

+    else:

+        rate = 22050

+

+    if args.local:

+        processor, model, tokenizer = load_local()

+    elif args.large:

+        processor, model, tokenizer, save_path = load_large()

+    else:

+        processor, model, tokenizer, save_path = load_model()

+

+    if args.save:

+        if not args.local:

+            save_model(processor, model, tokenizer, save_path)

+        else:

+            print("Local model is already saved, Its local...")

+        sys.exit(0)

+

+    if args.persistent:

+        persistent(rate, processor, model, tokenizer)

+        sys.exit(0)

+

+    if not args.text and not args.file:

+        print("No input text or file specified")

+        sys.exit(1)

+    elif args.text and args.file:

+        print("Both text and file specified")

+        sys.exit(1)

+

+    elif args.text:

+        speech = text_to_speech(processor, model, tokenizer, args.text)

+        if args.output:

+            output_file = args.output

+        else:

+            output_file = "output.wav"

+

+        write(output_file, rate, speech)

+

+    elif args.file:

+        speech = text_to_speech(processor, model, tokenizer, args.file)

+        if args.output:

+            output_file = args.output

+        else:

+            output_file = "output.wav"

+

+        write(output_file, 18000, speech)

+

+

+def persistent(rate, processor, model, tokenizer):

+

+    p = pyaudio.PyAudio()

+    stream = p.open(format=pyaudio.paFloat32, channels=1, rate=rate, output=True)

+

+    while True:

+        text = input("Enter text: ")

+        speech = text_to_speech(processor, model, tokenizer, text)

+        stream.write(speech.astype(np.float32).tobytes())

+

+    stream.stop_stream()

+    stream.close()

+    p.terminate()

+

+

+if __name__ == "__main__":

+    main()

diff --git a/setup.py b/setup.py

new file mode 100644

index 0000000..ad3e2a0

--- /dev/null

+++ b/setup.py

@@ -0,0 +1,24 @@

+from setuptools import setup, find_packages

+

+setup(

+    name='saywoof',

+    version='0.1.0',

+    description='An easy to use wrapper for the BARK text-to-speech engine',

+    author='Connor Etherington',

+    author_email='[email protected]',

+    packages=find_packages(),

+    install_requires=[

+        "pyaudio",

+        "numpy",

+        "soundfile",

+        "transformers",

+        "saywoof",

+        "colr",

+        "argparse"

+    ],

+    entry_points={

+        'console_scripts': [

+            'saywoof=saywoof.main:main'

+        ]

+    }

+)