sync

.gitattributes

@@ -1,35 +1,34 @@
 *.7z filter=lfs diff=lfs merge=lfs -text
 *.arrow filter=lfs diff=lfs merge=lfs -text
 *.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
 *.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ckpt filter=lfs diff=lfs merge=lfs -text
 *.ftz filter=lfs diff=lfs merge=lfs -text
 *.gz filter=lfs diff=lfs merge=lfs -text
 *.h5 filter=lfs diff=lfs merge=lfs -text
 *.joblib filter=lfs diff=lfs merge=lfs -text
 *.lfs.* filter=lfs diff=lfs merge=lfs -text
-*.mlmodel filter=lfs diff=lfs merge=lfs -text
 *.model filter=lfs diff=lfs merge=lfs -text
 *.msgpack filter=lfs diff=lfs merge=lfs -text
-*.npy filter=lfs diff=lfs merge=lfs -text
-*.npz filter=lfs diff=lfs merge=lfs -text
 *.onnx filter=lfs diff=lfs merge=lfs -text
 *.ot filter=lfs diff=lfs merge=lfs -text
 *.parquet filter=lfs diff=lfs merge=lfs -text
 *.pb filter=lfs diff=lfs merge=lfs -text
-*.pickle filter=lfs diff=lfs merge=lfs -text
-*.pkl filter=lfs diff=lfs merge=lfs -text
 *.pt filter=lfs diff=lfs merge=lfs -text
 *.pth filter=lfs diff=lfs merge=lfs -text
 *.rar filter=lfs diff=lfs merge=lfs -text
-*.safetensors filter=lfs diff=lfs merge=lfs -text
 saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tar filter=lfs diff=lfs merge=lfs -text
 *.tflite filter=lfs diff=lfs merge=lfs -text
 *.tgz filter=lfs diff=lfs merge=lfs -text
-*.wasm filter=lfs diff=lfs merge=lfs -text
 *.xz filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
-*.zst filter=lfs diff=lfs merge=lfs -text
-*tfevents* filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*.tfevents* filter=lfs diff=lfs merge=lfs -text
+*.db* filter=lfs diff=lfs merge=lfs -text
+*.ark* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,5 @@
+*.pt
+__pycache__/*
+tmp/*
+flagged/*
+rename.sh

README.md

@@ -1,13 +1,14 @@
 ---
 title: Pianos
-emoji: 🌖
+emoji: 🎹
 colorFrom: purple
 colorTo: red
 sdk: gradio
-sdk_version: 4.12.0
+sdk_version: 4.36.0
 app_file: app.py
 pinned: false
 license: mit
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Cite
+[1] <a href="https://arxiv.org/pdf/2310.04722.pdf">Monan Zhou, Shangda Wu, Shaohua Ji, Zijin Li, and Wei Li. A Holistic Evaluation of Piano Sound Quality[C]//Proceedings of the 10th Conference on Sound and Music Technology (CSMT). Springer, Singapore, 2023.</a>

app.py

@@ -0,0 +1,145 @@
+import os
+import torch
+import shutil
+import librosa
+import warnings
+import numpy as np
+import gradio as gr
+import librosa.display
+import matplotlib.pyplot as plt
+import torchvision.transforms as transforms
+from collections import Counter
+from PIL import Image
+from tqdm import tqdm
+from model import net, MODEL_DIR
+
+
+MODEL = net()
+
+
+def most_common_element(input_list):
+    counter = Counter(input_list)
+    mce, _ = counter.most_common(1)[0]
+    return mce
+
+
+def wav_to_mel(audio_path: str, width=0.18):
+    os.makedirs("./tmp")
+    try:
+        y, sr = librosa.load(audio_path, sr=48000)
+        non_silent = y
+        mel_spec = librosa.feature.melspectrogram(y=non_silent, sr=sr)
+        log_mel_spec = librosa.power_to_db(mel_spec, ref=np.max)
+        dur = librosa.get_duration(y=non_silent, sr=sr)
+        total_frames = log_mel_spec.shape[1]
+        step = int(width * total_frames / dur)
+        count = int(total_frames / step)
+        begin = int(0.5 * (total_frames - count * step))
+        end = begin + step * count
+        for i in tqdm(range(begin, end, step), desc="Converting wav to jpgs..."):
+            librosa.display.specshow(log_mel_spec[:, i : i + step])
+            plt.axis("off")
+            plt.savefig(
+                f"./tmp/{os.path.basename(audio_path)[:-4]}_{i}.jpg",
+                bbox_inches="tight",
+                pad_inches=0.0,
+            )
+            plt.close()
+
+    except Exception as e:
+        print(f"Error converting {audio_path} : {e}")
+
+
+def embed_img(img_path, input_size=224):
+    transform = transforms.Compose(
+        [
+            transforms.Resize([input_size, input_size]),
+            transforms.ToTensor(),
+            transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
+        ]
+    )
+    img = Image.open(img_path).convert("RGB")
+    return transform(img).unsqueeze(0)
+
+
+def inference(wav_path, folder_path="./tmp"):
+    if os.path.exists(folder_path):
+        shutil.rmtree(folder_path)
+
+    if not wav_path:
+        return None, "请输入音频 Please input an audio!"
+
+    wav_to_mel(wav_path)
+    outputs = []
+    all_files = os.listdir(folder_path)
+    for file_name in all_files:
+        if file_name.lower().endswith(".jpg"):
+            file_path = os.path.join(folder_path, file_name)
+            input = embed_img(file_path)
+            output: torch.Tensor = MODEL(input)
+            pred_id = torch.max(output.data, 1)[1]
+            outputs.append(pred_id)
+
+    max_count_item = most_common_element(outputs)
+    shutil.rmtree(folder_path)
+    return os.path.basename(wav_path), translate[classes[max_count_item]]
+
+
+if __name__ == "__main__":
+    warnings.filterwarnings("ignore")
+    translate = {
+        "PearlRiver": "Pearl River",
+        "YoungChang": "YOUNG CHANG",
+        "Steinway-T": "STEINWAY Theater",
+        "Hsinghai": "HSINGHAI",
+        "Kawai": "KAWAI",
+        "Steinway": "STEINWAY",
+        "Kawai-G": "KAWAI Grand",
+        "Yamaha": "YAMAHA",
+    }
+    classes = list(translate.keys())
+    example_wavs = []
+    for cls in classes:
+        example_wavs.append(f"{MODEL_DIR}/examples/{cls}.wav")
+
+    with gr.Blocks() as demo:
+        gr.Interface(
+            fn=inference,
+            inputs=gr.Audio(
+                type="filepath", label="上传钢琴录音 Upload a piano recording"
+            ),
+            outputs=[
+                gr.Textbox(label="音频文件名 Audio filename", show_copy_button=True),
+                gr.Textbox(
+                    label="钢琴分类结果 Piano classification result",
+                    show_copy_button=True,
+                ),
+            ],
+            examples=example_wavs,
+            cache_examples=False,
+            allow_flagging="never",
+            title="建议录音时长保持在 3s 左右, 过长会影响识别效率<br>It is recommended to keep the duration of recording around 3s, too long will affect the recognition efficiency.",
+        )
+
+        gr.Markdown(
+            """
+# 引用 Cite
+```bibtex
+@inproceedings{DBLP:journals/corr/abs-2310-04722,
+  author    = {Monan Zhou and
+               Shangda Wu and
+               Shaohua Ji and
+               Zijin Li and
+               Wei Li},
+  title     = {A Holistic Evaluation of Piano Sound Quality},
+  booktitle = {Proceedings of the 10th Conference on Sound and Music Technology (CSMT)},
+  year      = {2023},
+  publisher = {Springer Singapore},
+  address   = {Singapore},
+  timestamp = {Fri, 20 Oct 2023 12:04:38 +0200},
+  bibsource = {dblp computer science bibliography, https://dblp.org}
+}
+```"""
+        )
+
+    demo.launch()

model.py

@@ -0,0 +1,53 @@
+import torch
+import torch.nn as nn
+from modelscope import snapshot_download
+from torchvision.models import squeezenet1_1
+
+MODEL_DIR = snapshot_download(
+    "ccmusic-database/pianos",
+    cache_dir="./__pycache__",
+)
+
+
+def Classifier(cls_num=8, output_size=512, linear_output=False):
+    q = (1.0 * output_size / cls_num) ** 0.25
+    l1 = int(q * cls_num)
+    l2 = int(q * l1)
+    l3 = int(q * l2)
+
+    if linear_output:
+        return torch.nn.Sequential(
+            nn.Dropout(),
+            nn.Linear(output_size, l3),
+            nn.ReLU(inplace=True),
+            nn.Dropout(),
+            nn.Linear(l3, l2),
+            nn.ReLU(inplace=True),
+            nn.Dropout(),
+            nn.Linear(l2, l1),
+            nn.ReLU(inplace=True),
+            nn.Linear(l1, cls_num),
+        )
+
+    else:
+        return torch.nn.Sequential(
+            nn.Dropout(),
+            nn.Conv2d(output_size, l3, kernel_size=(1, 1), stride=(1, 1)),
+            nn.ReLU(inplace=True),
+            nn.AdaptiveAvgPool2d(output_size=(1, 1)),
+            nn.Flatten(),
+            nn.Linear(l3, l2),
+            nn.ReLU(inplace=True),
+            nn.Dropout(),
+            nn.Linear(l2, l1),
+            nn.ReLU(inplace=True),
+            nn.Linear(l1, cls_num),
+        )
+
+
+def net(weights=MODEL_DIR + "/save.pt"):
+    model = squeezenet1_1(pretrained=False)
+    model.classifier = Classifier()
+    model.load_state_dict(torch.load(weights, map_location=torch.device("cpu")))
+    model.eval()
+    return model

requirements.txt

@@ -0,0 +1,5 @@
+librosa
+torch
+matplotlib
+torchvision
+pillow