Phoneme Scorer V2 Wav2vec2

🚀 Wav2Vec2 LJSpeech Gruut

Wav2Vec2 LJSpeech Gruut is an automatic speech recognition model. It's based on the wav2vec 2.0 architecture and fine - tuned from Wav2Vec2 - Base on the LJSpech Phonemes dataset. Instead of predicting word sequences, it predicts phoneme sequences.

🚀 Quick Start

Prerequisites

Make sure you have installed the necessary libraries such as transformers, librosa, torch, and datasets.

Example Usage

from transformers import AutoProcessor, AutoModelForCTC, Wav2Vec2Processor
import librosa
import torch
from itertools import groupby
from datasets import load_dataset

def decode_phonemes(
    ids: torch.Tensor, processor: Wav2Vec2Processor, ignore_stress: bool = False
) -> str:
    """CTC-like decoding. First removes consecutive duplicates, then removes special tokens."""
    # removes consecutive duplicates
    ids = [id_ for id_, _ in groupby(ids)]

    special_token_ids = processor.tokenizer.all_special_ids + [
        processor.tokenizer.word_delimiter_token_id
    ]
    # converts id to token, skipping special tokens
    phonemes = [processor.decode(id_) for id_ in ids if id_ not in special_token_ids]

    # joins phonemes
    prediction = " ".join(phonemes)

    # whether to ignore IPA stress marks
    if ignore_stress == True:
        prediction = prediction.replace("ˈ", "").replace("ˌ", "")

    return prediction

checkpoint = "bookbot/wav2vec2-ljspeech-gruut"

model = AutoModelForCTC.from_pretrained(checkpoint)
processor = AutoProcessor.from_pretrained(checkpoint)
sr = processor.feature_extractor.sampling_rate

# load dummy dataset and read soundfiles
ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")
audio_array = ds[0]["audio"]["array"]

# or, read a single audio file
# audio_array, _ = librosa.load("myaudio.wav", sr=sr)

inputs = processor(audio_array, return_tensors="pt", padding=True)

with torch.no_grad():
    logits = model(inputs["input_values"]).logits

predicted_ids = torch.argmax(logits, dim=-1)
prediction = decode_phonemes(predicted_ids[0], processor, ignore_stress=True)
# => should give 'b ɪ k ʌ z j u ɚ z s l i p ɪ ŋ ɪ n s t ɛ d ə v k ɔ ŋ k ɚ ɪ ŋ ð ə l ʌ v l i ɹ z p ɹ ɪ n s ə s h æ z b ɪ k ʌ m ə v f ɪ t ə l w ɪ θ n b oʊ p ɹ ə ʃ æ ɡ i s ɪ t s ð ɛ ɹ ə k u ɪ ŋ d ʌ v'

✨ Features

• Phoneme Prediction: Trained to predict sequences of phonemes instead of words.
• Fine - Tuned Model: Based on the Wav2Vec2 - Base model, fine - tuned on the LJSpeech Phonemes dataset.

📦 Installation

The installation steps are not provided in the original README. You can install the required libraries using pip or conda according to the usage example. For example:

pip install transformers librosa torch datasets

💻 Usage Examples

Basic Usage

# The basic usage code is the same as the quick start example
from transformers import AutoProcessor, AutoModelForCTC, Wav2Vec2Processor
import librosa
import torch
from itertools import groupby
from datasets import load_dataset

def decode_phonemes(
    ids: torch.Tensor, processor: Wav2Vec2Processor, ignore_stress: bool = False
) -> str:
    """CTC-like decoding. First removes consecutive duplicates, then removes special tokens."""
    # removes consecutive duplicates
    ids = [id_ for id_, _ in groupby(ids)]

    special_token_ids = processor.tokenizer.all_special_ids + [
        processor.tokenizer.word_delimiter_token_id
    ]
    # converts id to token, skipping special tokens
    phonemes = [processor.decode(id_) for id_ in ids if id_ not in special_token_ids]

    # joins phonemes
    prediction = " ".join(phonemes)

    # whether to ignore IPA stress marks
    if ignore_stress == True:
        prediction = prediction.replace("ˈ", "").replace("ˌ", "")

    return prediction

checkpoint = "bookbot/wav2vec2-ljspeech-gruut"

model = AutoModelForCTC.from_pretrained(checkpoint)
processor = AutoProcessor.from_pretrained(checkpoint)
sr = processor.feature_extractor.sampling_rate

# load dummy dataset and read soundfiles
ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")
audio_array = ds[0]["audio"]["array"]

# or, read a single audio file
# audio_array, _ = librosa.load("myaudio.wav", sr=sr)

inputs = processor(audio_array, return_tensors="pt", padding=True)

with torch.no_grad():
    logits = model(inputs["input_values"]).logits

predicted_ids = torch.argmax(logits, dim=-1)
prediction = decode_phonemes(predicted_ids[0], processor, ignore_stress=True)

Advanced Usage

There is no advanced usage example in the original README. You can further explore the model by adjusting the hyperparameters during training or using different datasets for inference.

📚 Documentation

Model Information

Evaluation Results

🔧 Technical Details

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 30.0
• mixed_precision_training: Native AMP

Training results

📄 License

This model is licensed under the Apache 2.0 license.

⚠️ Important Note

Do consider the biases which came from pre - training datasets that may be carried over into the results of this model.

👥 Authors

Wav2Vec2 LJSpeech Gruut was trained and evaluated by Wilson Wongso. All computation and development are done on Google Cloud.

🛠️ Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.10.0
• Datasets 2.7.1
• Tokenizers 0.13.2
• Gruut 2.3.4