clean up README/notebooks, reduce to 3 epochs, update JSON

Author: alexsin368

AI-and-Analytics/End-to-end-Workloads/LanguageIdentification/Inference/lang_id_inference.ipynb

@@ -132,7 +132,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!python quantize_model.py -p ./lang_id_commonvoice_model -datapath $COMMON_VOICE_PATH/dev"
+    "!python quantize_model.py -p ./lang_id_commonvoice_model -datapath $COMMON_VOICE_PATH/processed_data/dev"
    ]
   },
   {

AI-and-Analytics/End-to-end-Workloads/LanguageIdentification/README.md

@@ -6,7 +6,7 @@ Languages are selected from the CommonVoice dataset for training, validation, an
 
 | Area                  | Description
 |:---                   |:---
-| What you will learn   | How to use training and inference with SpeechBrain, Intel® Extension for PyTorch (IPEX) inference, Intel® Neural Compressor (INC) quantization, and a oneapi-aikit container
+| What you will learn   | How to use training and inference with SpeechBrain, Intel® Extension for PyTorch* (IPEX) inference, Intel® Neural Compressor (INC) quantization
 | Time to complete      | 60 minutes
 
 ## Purpose
@@ -18,8 +18,8 @@ Spoken audio comes in different languages and this sample uses a model to identi
 | Optimized for        | Description
 |:---                  |:---
 | OS                   | Ubuntu* 22.04 or newer
-| Hardware             | Intel® Xeon® processor family
-| Software             | Intel® OneAPI AI Analytics Toolkit <br> Hugging Face SpeechBrain
+| Hardware             | Intel® Xeon® and Core® processor families
+| Software             | Intel® AI Tools <br> Hugging Face SpeechBrain
 
 ## Key Implementation Details
 
@@ -41,15 +41,14 @@ For both training and inference, you can run the sample and scripts in Jupyter N
 
 1. Create your conda environment by following the instructions on the Intel [AI Tools Selector](https://www.intel.com/content/www/us/en/developer/tools/oneapi/ai-tools-selector.html). You can follow these settings:
 
-* AI Tools
-* Preset: Inference Optimization
-* Distribution Type: conda*
+* Tool: AI Tools
+* Preset or customize: Customize
+* Distribution Type: conda* or pip
 * Python Versions: Python* 3.9 or 3.10
+* PyTorch* Framework Optimizations: Intel® Extension for PyTorch* (CPU)
+* Intel®-Optimized Tools & Libraries: Intel® Neural Compressor
 
-Then activate your environment:
-```bash
-conda activate <your-env-name>
-```
+>**Note**: Be sure to activate your environment before installing the packages. If using pip, install using `python -m pip` instead of just `pip`.
 
 2. Create your dataset folder and set the environment variable `COMMON_VOICE_PATH`. This needs to match with where you downloaded your dataset.
 ```bash
@@ -221,7 +220,7 @@ After training, the output should be inside the `results/epaca/1987` folder. By
 
    cp classifier.ckpt ../../.
    cp embedding_model.ckpt ../../
-   cd ../..
+   cd ../../../..
    ```
 
    You may need to modify the permissions of these files to be executable i.e. `sudo chmod 755` before you run the inference scripts to consume them.

AI-and-Analytics/End-to-end-Workloads/LanguageIdentification/Training/clean.sh

@@ -1,4 +1,4 @@
 #!/bin/bash
 
 echo "Deleting rir, noise, speechbrain"
-rm -R rir noise speechbrain
+rm -R rir noise

AI-and-Analytics/End-to-end-Workloads/LanguageIdentification/Training/lang_id_training.ipynb

@@ -29,9 +29,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!cp speechbrain/recipes/VoxLingua107/lang_id/create_wds_shards.py create_wds_shards.py\n",
-    "!cp speechbrain/recipes/VoxLingua107/lang_id/train.py train.py\n",
-    "!cp speechbrain/recipes/VoxLingua107/lang_id/hparams/train_ecapa.yaml train_ecapa.yaml"
+    "!cp ../speechbrain/recipes/VoxLingua107/lang_id/create_wds_shards.py create_wds_shards.py\n",
+    "!cp ../speechbrain/recipes/VoxLingua107/lang_id/train.py train.py\n",
+    "!cp ../speechbrain/recipes/VoxLingua107/lang_id/hparams/train_ecapa.yaml train_ecapa.yaml"
    ]
   },
   {
@@ -166,21 +166,26 @@
    "metadata": {},
    "outputs": [],
    "source": [
+    "import os\n",
+    "\n",
     "# 1)\n",
     "!cp -R results/epaca/1987 ../Inference/lang_id_commonvoice_model\n",
     "\n",
     "# 2)\n",
-    "!cd ../Inference/lang_id_commonvoice_model/save\n",
+    "os.chdir(\"../Inference/lang_id_commonvoice_model/save\")\n",
     "\n",
     "# 3)\n",
     "!cp label_encoder.txt ../.\n",
     "\n",
-    "# 4)\n",
-    "# Navigate into the CKPT folder\n",
-    "!cd CKPT* # Set this to your CKPT folder. By default it will navigate into the one that is present.\n",
+    "# 4) \n",
+    "folders = os.listdir()\n",
+    "for folder in folders:\n",
+    "    if \"CKPT\" in folder:\n",
+    "        os.chdir(folder)\n",
+    "        break\n",
     "!cp classifier.ckpt ../../.\n",
     "!cp embedding_model.ckpt ../../\n",
-    "!cd ../.."
+    "os.chdir(\"../../../..\")"
    ]
   },
   {

AI-and-Analytics/End-to-end-Workloads/LanguageIdentification/Training/train_ecapa.patch

@@ -27,7 +27,7 @@
 
  # Training parameters
 -number_of_epochs: 40
-+number_of_epochs: 10
++number_of_epochs: 3
  lr: 0.001
  lr_final: 0.0001
  sample_rate: 16000

AI-and-Analytics/End-to-end-Workloads/LanguageIdentification/sample.json

@@ -15,6 +15,7 @@
           "export COMMON_VOICE_PATH=/data/commonVoice"
         ],
         "steps": [
+          "mkdir -p /data/commonVoice",
           "apt-get update && apt-get install ffmpeg libgl1 -y",
           "source initialize.sh",
           "cd ./Dataset",