File size: 3,709 Bytes
381d0aa |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 |
---
license: mit
---
## Mel-Spectrogram Image Dataset (Generated via Custom Pipeline)
> **This dataset was fully generated through my notebook
> *“Building an Audio Classification Pipeline with DL”* available on my profile.**
> It represents a complete end-to-end transformation from raw audio to clean, balanced Mel-spectrogram images suitable for deep learning.
---
### **Dataset Summary**
| Property | Description |
| ---------------------------- | --------------------------------------------- |
| **Number of Classes** | 13 distinct audio categories |
| **Original Audio per Class** | ~40 raw recordings |
| **Average Duration** | ~5 seconds per audio file |
| **Final Images per Class** | 125 Mel-spectrogram images |
| **Final Dataset Size** | 13 × 125 = **1625 images** |
| **Sampling Rate** | Standardized to **16 kHz** |
| **Audio Length** | Uniform **5-second** fixed length |
| **Spectrogram Type** | 128-Mel frequency bins, `melspectrogram → dB` |
---
### **High-Level Processing Pipeline**
The dataset was built using a **fully custom preprocessing, cleaning, and augmentation pipeline**, implemented step-by-step in the notebook.
#### **1. Data Ingestion**
* Loaded all raw audio files from 13 folders
* Parsed metadata (sample rate, duration, amplitude, SNR, etc.)
#### **2. Cleaning & Standardization**
* Removed corrupt, silent, or unreadable audio files
* Normalized peak amplitudes
* Trimmed silence using `librosa.effects.trim`
* Performed noise reduction (`noisereduce`)
* Converted all audio to **mono**
* Resampled to **16,000 Hz**
* Ensured each sample is **exactly 5 seconds**
#### **3. Dataset Balancing**
* Used augmentation for minority classes
* Used controlled undersampling or oversampling where necessary
* Verified all classes contain equal counts
#### **4. Audio Augmentation (Used for Balancing & Variability)**
Augmentations built with **audiomentations**:
* Time shift
* Pitch shift
* Time stretching
* Gaussian noise injection
* Random perturbations for robustness
#### **5. Splitting & Chunking**
* Long samples were split into 5-second chunks
* Shorter samples padded to match target duration
* Ensured strict uniformity before feature extraction
#### **6. Mel-Spectrogram Generation**
Converted all cleaned audio files into Mel-spectrogram images using:
* `n_fft = 1024`
* `hop_length = 512`
* `n_mels = 128`
* Converted to decibel scale (`power_to_db`)
* Saved images in **RGBA format** to preserve color-mapped spectral information
---
### **Final Technical Description**
> **“The final dataset consists of 13 audio classes, each expanded to exactly 125 Mel-spectrogram images through a rigorous pipeline of cleaning, normalization, augmentation, noise reduction, resampling, duration standardization, and feature extraction. All processing steps were implemented in my notebook *‘Building an Audio Classification Pipeline with DL,’* where raw 5-second audio recordings were transformed into high-quality Mel-spectrogram images suitable for deep learning models.”**
---
### **Examples of the Images**
.png?generation=1763570855911665&alt=media) |