[cli] add first cli support

README.md

@@ -16,6 +16,29 @@ online feature extraction or loading pre-extracted features in kaldi-format.
 
 ## Installation
 
+### Install python package
+``` sh
+pip install git+https://github.com/wenet-e2e/wespeaker.git
+```
+**Command-line usage** (use `-h` for parameters):
+
+``` sh
+$ wespeaker --task embedding --audio_file audio.wav
+$ wespeaker --task similarity --audio_file audio.wav --audio_file2 audio2.wav
+$ wespeaker --task diarization --audio_file audio.wav # TODO
+```
+
+**Python programming usage**:
+
+``` python
+import wespeaker
+
+model = wespeaker.load_model('chinese')
+embedding = model.extract_embeding('audio.wav')
+similarity = model.compute_similarity('audio1.wav', 'audio2.wav')
+```
+
+### Install for development & deployment
 * Clone this repo
 ``` sh
 git clone https://github.com/wenet-e2e/wespeaker.git

setup.cfg

@@ -0,0 +1,15 @@
+[metadata]
+name = wespeaker
+version = 0.0.0
+license = Apache Software License
+description = End to end speech speaker toolkit
+long_description = file: README.md
+classifiers =
+    License :: OSI Approved :: Apache Software License
+    Operating System :: OS Independent
+    Programming Language :: Python :: 3
+
+[options]
+packages = find:
+include_package_data = True
+python_requires = >= 3.8

setup.py

@@ -0,0 +1,17 @@
+from setuptools import setup, find_packages
+
+requirements = [
+    "tqdm",
+    "onnxruntime>=1.12.0",
+    "python-speech-features>=0.6",
+    "scipy>=1.5.2",
+]
+
+setup(
+    name="wespeaker",
+    install_requires=requirements,
+    packages=find_packages(),
+    entry_points={"console_scripts": [
+        "wespeaker = wespeaker.cli.speaker:main",
+    ]},
+)

wespeaker/__init__.py

@@ -0,0 +1 @@
+from wespeaker.cli.speaker import load_model  # noqa

wespeaker/cli/hub.py

@@ -0,0 +1,79 @@
+# Copyright (c) 2022  Mddct([email protected])
+#               2023  Binbin Zhang([email protected])
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import requests
+import sys
+from pathlib import Path
+from urllib.request import urlretrieve
+
+import tqdm
+
+
+def download(url: str, dest: str):
+    """ download from url to dest
+    """
+    print('Downloading {} to {}'.format(url, dest))
+
+    def progress_hook(t):
+        last_b = [0]
+
+        def update_to(b=1, bsize=1, tsize=None):
+            if tsize not in (None, -1):
+                t.total = tsize
+            displayed = t.update((b - last_b[0]) * bsize)
+            last_b[0] = b
+            return displayed
+
+        return update_to
+
+    # *.tar.gz
+    name = url.split('?')[0].split('/')[-1]
+    with tqdm.tqdm(unit='B',
+                   unit_scale=True,
+                   unit_divisor=1024,
+                   miniters=1,
+                   desc=(name)) as t:
+        urlretrieve(url, filename=dest, reporthook=progress_hook(t), data=None)
+        t.total = t.n
+
+
+class Hub(object):
+    Assets = {
+        "chinese": "cnceleb_resnet34.onnx",
+        "english": "voxceleb_resnet221_LM.onnx",
+    }
+
+    def __init__(self) -> None:
+        pass
+
+    @staticmethod
+    def get_model(lang: str) -> str:
+        if lang not in Hub.Assets.keys():
+            print('ERROR: Unsupported lang {} !!!'.format(lang))
+            sys.exit(1)
+        model = Hub.Assets[lang]
+        model_path = os.path.join(Path.home(), ".wespeaker", model)
+        if not os.path.exists(model_path):
+            if not os.path.exists(os.path.dirname(model_path)):
+                os.makedirs(os.path.dirname(model_path))
+            response = requests.get(
+                "https://modelscope.cn/api/v1/datasets/wenet/wespeaker_pretrained_models/oss/tree"  # noqa
+            )
+            model_info = next(data for data in response.json()["Data"]
+                              if data["Key"] == model)
+            model_url = model_info['Url']
+            download(model_url, model_path)
+        return model_path

wespeaker/cli/speaker.py

@@ -0,0 +1,107 @@
+# Copyright (c) 2023 Binbin Zhang ([email protected])
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import argparse
+import sys
+
+import numpy as np
+import onnxruntime as ort
+import scipy.io.wavfile as wav
+from numpy.linalg import norm
+from python_speech_features import fbank
+
+from wespeaker.cli.hub import Hub
+
+
+class Speaker:
+    def __init__(self, model_path: str):
+        self.session = ort.InferenceSession(model_path)
+
+    def extract_embedding(self, audio_path: str):
+        sample_rate, pcm = wav.read(audio_path)
+        # TODO(Binbin Zhang): verify the feat
+        feats, _ = fbank(pcm,
+                         sample_rate,
+                         nfilt=80,
+                         lowfreq=20,
+                         winfunc=np.hamming)
+        feats = np.log(feats)
+        feats = np.expand_dims(feats, axis=0).astype(np.float32)
+        outputs = self.session.run(None, {"feats": feats})
+        embedding = outputs[0][0]
+        return embedding
+
+    def compute_similarity(self, audio_path1: str, audio_path2) -> float:
+        e1 = self.extract_embedding(audio_path1)
+        e2 = self.extract_embedding(audio_path2)
+        s = np.dot(e1, e2) / (norm(e1) * norm(e2))
+        return s
+
+    # TODO(Chengdong Liang): Add implementation
+    def register(self, audio_path: str):
+        pass
+
+    # TODO(Chengdong Liang): Add implementation
+    def recognize(self, audio_path: str):
+        pass
+
+
+def load_model(language: str) -> Speaker:
+    model_path = Hub.get_model(language)
+    return Speaker(model_path)
+
+
+def get_args():
+    parser = argparse.ArgumentParser(description='')
+    parser.add_argument('-t',
+                        '--task',
+                        choices=[
+                            'embedding',
+                            'similarity',
+                            'diarization',
+                        ],
+                        default='embedding',
+                        help='task type')
+    parser.add_argument('-l',
+                        '--language',
+                        choices=[
+                            'chinese',
+                            'english',
+                        ],
+                        default='chinese',
+                        help='language type')
+    parser.add_argument('--audio_file', help='audio file')
+    parser.add_argument('--audio_file2',
+                        help='audio file2, for similarity task')
+    args = parser.parse_args()
+    return args
+
+
+def main():
+    args = get_args()
+    model = load_model(args.language)
+    if args.task == 'embedding':
+        print(model.extract_embedding(args.audio_file))
+    elif args.task == 'similarity':
+        print(model.compute_similarity(args.audio_file, args.audio_file2))
+    elif args.task == 'diarization':
+        # TODO(Chengdong Liang): Add diarization surport
+        pass
+    else:
+        print('Unsupported task {}'.format(args.task))
+        sys.exit(-1)
+
+
+if __name__ == '__main__':
+    main()