Subindo arquivos

README.md

@@ -1,13 +1,65 @@
 ---
-title: Speech Accent Pt Br Classifier
-emoji: 📚
+title: Speech-accent-pt-br-classifier
+emoji: 🎙️🤖🇧🇷
 colorFrom: blue
-colorTo: indigo
+colorTo: green
 sdk: gradio
-sdk_version: 4.36.1
+sdk_version: "4.12.0"
 app_file: app.py
 pinned: false
-license: ecl-2.0
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Speech Portuguese (Brazilian) Accent Classifier
+
+This project is a speech accent classifier that distinguishes between Portuguese (Brazilian) and other accents.
+
+## Project Overview
+
+This application uses a trained model to classify speech accents into two categories:
+1. Portuguese (Brazilian)
+2. Other
+
+The model is based on the author's work [(results) brazil pt accent] and utilizes the Portuguese portion of the Common Voice dataset (version 11.0) from Mozilla Foundation.
+
+## Dataset
+
+The project uses the Portuguese subset of the Common Voice dataset:
+- Dataset: "mozilla-foundation/common_voice_11_0", "pt"
+
+Brazilian accents included in the dataset:
+- Português do Brasil, Região Sul do Brasil
+- Paulistano
+- Paulista, Brasileiro
+- Carioca
+- Mato Grosso
+- Mineiro
+- Interior Paulista
+- Gaúcho
+- Nordestino
+- And various regional mixes
+
+## Technical Details
+
+The project utilizes the following model and processor:
+- Model: "facebook/wav2vec2-base-960h"
+- Processor: Wav2Vec2Processor.from_pretrained
+
+## License
+
+ecl
+
+## Developer Information
+
+Developed by Ramon Mayor Martins (2024)
+- Email: rmayormartins@gmail.com
+- Homepage: https://rmayormartins.github.io/
+- Twitter: @rmayormartins
+- GitHub: https://github.com/rmayormartins
+
+## Acknowledgements
+
+Special thanks to Instituto Federal de Santa Catarina (Federal Institute of Santa Catarina) IFSC-São José-Brazil.
+
+## Contact
+
+For any queries or suggestions, please contact the developer using the information provided above.

app.py

@@ -0,0 +1,50 @@
+import gradio as gr
+import torch
+import numpy as np
+from transformers import Wav2Vec2Processor, Wav2Vec2ForSequenceClassification
+
+# modelo e o processador salvos
+model_name = "results"
+processor = Wav2Vec2Processor.from_pretrained(model_name)
+model = Wav2Vec2ForSequenceClassification.from_pretrained(model_name)
+
+def classify_accent(audio):
+    if audio is None:
+        return "Erro: Nenhum áudio recebido"
+
+    # entrada
+    print(f"Tipo de entrada de áudio: {type(audio)}")
+
+    # O áudio formato 
+    print(f"Received audio input: {audio}")
+
+    try:
+        audio_array = audio[1]  # O áudio da tupla
+        sample_rate = audio[0]  # A taxa de amostragem da tupla
+
+        print(f"Shape do áudio: {audio_array.shape}, Taxa de amostragem: {sample_rate}")
+
+        # 
+        audio_array = audio_array.astype(np.float32)
+
+        # taxa de amostragem
+        if sample_rate != 16000:
+            import librosa
+            audio_array = librosa.resample(audio_array, orig_sr=sample_rate, target_sr=16000)
+
+        input_values = processor(audio_array, return_tensors="pt", sampling_rate=16000).input_values
+        # Inf
+        with torch.no_grad():
+            logits = model(input_values).logits
+        predicted_ids = torch.argmax(logits, dim=-1).item()
+
+        # ids accent
+        labels = ["Brazilian", "Outro"]
+        return labels[predicted_ids]
+
+    except Exception as e:
+        return f"Erro ao processar o áudio: {str(e)}"
+
+# Interface do Gradio
+interface = gr.Interface(fn=classify_accent, inputs=gr.Audio(type="numpy"), outputs="label")
+interface.launch()

requirements.txt

@@ -0,0 +1,5 @@
+gradio==4.29.0
+torch
+transformers
+librosa
+numpy

results/config.json

@@ -0,0 +1,110 @@
+{
+  "_name_or_path": "facebook/wav2vec2-base-960h",
+  "activation_dropout": 0.1,
+  "adapter_attn_dim": null,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForSequenceClassification"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "codevector_dim": 256,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": false,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "sum",
+  "ctc_zero_infinity": false,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": false,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "group",
+  "feat_proj_dropout": 0.1,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.1,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.1,
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.1,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "model_type": "wav2vec2",
+  "num_adapter_layers": 3,
+  "num_attention_heads": 12,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 12,
+  "num_negatives": 100,
+  "output_hidden_size": 768,
+  "pad_token_id": 0,
+  "proj_codevector_dim": 256,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "torch_dtype": "float32",
+  "transformers_version": "4.41.2",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 32,
+  "xvector_output_dim": 512
+}

results/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2832c855da20b63ee6353ba032d27bd9c5c4920c64bc1fe49fe5d498f6f87d0
+size 378302360

results/preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "processor_class": "Wav2Vec2Processor",
+  "return_attention_mask": false,
+  "sampling_rate": 16000
+}

results/special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

results/tokenizer_config.json

@@ -0,0 +1,50 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": true,
+  "do_lower_case": false,
+  "do_normalize": true,
+  "eos_token": "</s>",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "processor_class": "Wav2Vec2Processor",
+  "replace_word_delimiter_char": " ",
+  "return_attention_mask": false,
+  "target_lang": null,
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "unk_token": "<unk>",
+  "word_delimiter_token": "|"
+}

results/vocab.json

@@ -0,0 +1,34 @@
+{
+  "'": 27,
+  "</s>": 2,
+  "<pad>": 0,
+  "<s>": 1,
+  "<unk>": 3,
+  "A": 7,
+  "B": 24,
+  "C": 19,
+  "D": 14,
+  "E": 5,
+  "F": 20,
+  "G": 21,
+  "H": 11,
+  "I": 10,
+  "J": 29,
+  "K": 26,
+  "L": 15,
+  "M": 17,
+  "N": 9,
+  "O": 8,
+  "P": 23,
+  "Q": 30,
+  "R": 13,
+  "S": 12,
+  "T": 6,
+  "U": 16,
+  "V": 25,
+  "W": 18,
+  "X": 28,
+  "Y": 22,
+  "Z": 31,
+  "|": 4
+}