app.py

#!/usr/bin/env python3
import sys
import torch
import logging
import gradio as gr
import speechbrain as sb
from pathlib import Path
import os
import torchaudio
from hyperpyyaml import load_hyperpyyaml
from speechbrain.tokenizers.SentencePiece import SentencePiece
from speechbrain.utils.data_utils import undo_padding
from speechbrain.utils.distributed import run_on_main

"""Recipe for training a sequence-to-sequence ASR system with CommonVoice.
The system employs a wav2vec2 encoder and a CTC decoder.
Decoding is performed with greedy decoding (will be extended to beam search).

To run this recipe, do the following:
> python train_with_wav2vec2.py hparams/train_with_wav2vec2.yaml

With the default hyperparameters, the system employs a pretrained wav2vec2 encoder.
The wav2vec2 model is pretrained following the model given in the hprams file.
It may be dependent on the language.

The neural network is trained with CTC on sub-word units estimated with
Byte Pairwise Encoding (BPE).

The experiment file is flexible enough to support a large variety of
different systems. By properly changing the parameter files, you can try
different encoders, decoders, tokens (e.g, characters instead of BPE),
training languages (all CommonVoice languages), and many
other possible variations.

Authors
 * Titouan Parcollet 2021
"""

logger = logging.getLogger(__name__)


# Define training procedure
class ASR(sb.core.Brain):
    def compute_forward(self, batch, stage):
        """Forward computations from the waveform batches to the output probabilities."""

        batch = batch.to(self.device)
        wavs, wav_lens = batch.sig
        wavs, wav_lens = wavs.to(self.device), wav_lens.to(self.device)
        if stage == sb.Stage.TRAIN:
            if hasattr(self.hparams, "augmentation"):
                wavs = self.hparams.augmentation(wavs, wav_lens)

        # Forward pass
        feats = self.modules.wav2vec2(wavs, wav_lens)
        x = self.modules.enc(feats)
        logits = self.modules.ctc_lin(x)
        p_ctc = self.hparams.log_softmax(logits)

        return p_ctc, wav_lens
    
    def treat_wav(self,sig):
        feats = self.modules.wav2vec2(sig.to("cpu"), torch.tensor([1]).to("cpu"))
        feats = self.modules.enc(feats)
        logits = self.modules.ctc_lin(feats)
        p_ctc = self.hparams.log_softmax(logits)
        predicted_words =[]
        for logs in p_ctc:
            text = decoder.decode(logs.detach().cpu().numpy())
            predicted_words.append(text.split(" "))
        return " ".join(predicted_words[0])
 
    def compute_objectives(self, predictions, batch, stage):
        """Computes the loss (CTC) given predictions and targets."""

        p_ctc, wav_lens = predictions

        ids = batch.id
        tokens, tokens_lens = batch.tokens

        loss = self.hparams.ctc_cost(p_ctc, tokens, wav_lens, tokens_lens)

        if stage != sb.Stage.TRAIN:
            predicted_tokens = sb.decoders.ctc_greedy_decode(
                p_ctc, wav_lens, blank_id=self.hparams.blank_index
            )
            # Decode token terms to words
            if self.hparams.use_language_modelling:
                predicted_words = []
                for logs in p_ctc:
                    text = decoder.decode(logs.detach().cpu().numpy())
                    predicted_words.append(text.split(" "))
            else:
                predicted_words = [
                    "".join(self.tokenizer.decode_ndim(utt_seq)).split(" ")
                    for utt_seq in predicted_tokens
                ]
            # Convert indices to words
            target_words = [wrd.split(" ") for wrd in batch.wrd]

            self.wer_metric.append(ids, predicted_words, target_words)
            self.cer_metric.append(ids, predicted_words, target_words)

        return loss

    def fit_batch(self, batch):
        """Train the parameters given a single batch in input"""
        should_step = self.step % self.grad_accumulation_factor == 0
        # Managing automatic mixed precision
        # TOFIX: CTC fine-tuning currently is unstable
        # This is certainly due to CTC being done in fp16 instead of fp32
        if self.auto_mix_prec:
            with torch.cuda.amp.autocast():
                with self.no_sync():
                    outputs = self.compute_forward(batch, sb.Stage.TRAIN)
                loss = self.compute_objectives(outputs, batch, sb.Stage.TRAIN)
            with self.no_sync(not should_step):
                self.scaler.scale(
                    loss / self.grad_accumulation_factor
                ).backward()
            if should_step:

                if not self.hparams.wav2vec2.freeze:
                    self.scaler.unscale_(self.wav2vec_optimizer)
                self.scaler.unscale_(self.model_optimizer)
                if self.check_gradients(loss):
                    if not self.hparams.wav2vec2.freeze:
                        if self.optimizer_step >= self.hparams.warmup_steps:
                            self.scaler.step(self.wav2vec_optimizer)
                    self.scaler.step(self.model_optimizer)
                self.scaler.update()
                self.zero_grad()
                self.optimizer_step += 1
        else:
            # This is mandatory because HF models have a weird behavior with DDP
            # on the forward pass
            with self.no_sync():
                outputs = self.compute_forward(batch, sb.Stage.TRAIN)

            loss = self.compute_objectives(outputs, batch, sb.Stage.TRAIN)

            with self.no_sync(not should_step):
                (loss / self.grad_accumulation_factor).backward()
            if should_step:
                if self.check_gradients(loss):
                    if not self.hparams.wav2vec2.freeze:
                        if self.optimizer_step >= self.hparams.warmup_steps:
                            self.wav2vec_optimizer.step()
                    self.model_optimizer.step()
                self.zero_grad()
                self.optimizer_step += 1

        self.on_fit_batch_end(batch, outputs, loss, should_step)
        return loss.detach().cpu()

    def evaluate_batch(self, batch, stage):
        """Computations needed for validation/test batches"""
        predictions = self.compute_forward(batch, stage=stage)
        with torch.no_grad():
            loss = self.compute_objectives(predictions, batch, stage=stage)
        return loss.detach()

    def on_stage_start(self, stage, epoch):
        """Gets called at the beginning of each epoch"""
        if stage != sb.Stage.TRAIN:
            self.cer_metric = self.hparams.cer_computer()
            self.wer_metric = self.hparams.error_rate_computer()

    def on_stage_end(self, stage, stage_loss, epoch):
        """Gets called at the end of an epoch."""
        # Compute/store important stats
        stage_stats = {"loss": stage_loss}
        if stage == sb.Stage.TRAIN:
            self.train_stats = stage_stats
        else:
            stage_stats["CER"] = self.cer_metric.summarize("error_rate")
            stage_stats["WER"] = self.wer_metric.summarize("error_rate")

        # Perform end-of-iteration things, like annealing, logging, etc.
        if stage == sb.Stage.VALID:
            old_lr_model, new_lr_model = self.hparams.lr_annealing_model(
                stage_stats["loss"]
            )
            old_lr_wav2vec, new_lr_wav2vec = self.hparams.lr_annealing_wav2vec(
                stage_stats["loss"]
            )
            sb.nnet.schedulers.update_learning_rate(
                self.model_optimizer, new_lr_model
            )
            if not self.hparams.wav2vec2.freeze:
                sb.nnet.schedulers.update_learning_rate(
                    self.wav2vec_optimizer, new_lr_wav2vec
                )
            self.hparams.train_logger.log_stats(
                stats_meta={
                    "epoch": epoch,
                    "lr_model": old_lr_model,
                    "lr_wav2vec": old_lr_wav2vec,
                },
                train_stats=self.train_stats,
                valid_stats=stage_stats,
            )
            self.checkpointer.save_and_keep_only(
                meta={"WER": stage_stats["WER"]}, min_keys=["WER"],
            )
        elif stage == sb.Stage.TEST:
            self.hparams.train_logger.log_stats(
                stats_meta={"Epoch loaded": self.hparams.epoch_counter.current},
                test_stats=stage_stats,
            )
            with open(self.hparams.wer_file, "w") as w:
                self.wer_metric.write_stats(w)

    def init_optimizers(self):
        "Initializes the wav2vec2 optimizer and model optimizer"

        # If the wav2vec encoder is unfrozen, we create the optimizer
        if not self.hparams.wav2vec2.freeze:
            self.wav2vec_optimizer = self.hparams.wav2vec_opt_class(
                self.modules.wav2vec2.parameters()
            )
            if self.checkpointer is not None:
                self.checkpointer.add_recoverable(
                    "wav2vec_opt", self.wav2vec_optimizer
                )

        self.model_optimizer = self.hparams.model_opt_class(
            self.hparams.model.parameters()
        )

        if self.checkpointer is not None:
            self.checkpointer.add_recoverable("modelopt", self.model_optimizer)

    def zero_grad(self, set_to_none=False):
        if not self.hparams.wav2vec2.freeze:
            self.wav2vec_optimizer.zero_grad(set_to_none)
        self.model_optimizer.zero_grad(set_to_none)


# Define custom data procedure
def dataio_prepare(hparams):
    """This function prepares the datasets to be used in the brain class.
    It also defines the data processing pipeline through user-defined functions."""

    # 1. Define datasets
    data_folder = hparams["data_folder"]

    train_data = sb.dataio.dataset.DynamicItemDataset.from_csv(
        csv_path=hparams["train_csv"], replacements={"data_root": data_folder},
    )

    if hparams["sorting"] == "ascending":
        # we sort training data to speed up training and get better results.
        train_data = train_data.filtered_sorted(
            sort_key="duration",
            key_max_value={"duration": hparams["avoid_if_longer_than"]},
        )
        # when sorting do not shuffle in dataloader ! otherwise is pointless
        hparams["dataloader_options"]["shuffle"] = False

    elif hparams["sorting"] == "descending":
        train_data = train_data.filtered_sorted(
            sort_key="duration",
            reverse=True,
            key_max_value={"duration": hparams["avoid_if_longer_than"]},
        )
        # when sorting do not shuffle in dataloader ! otherwise is pointless
        hparams["dataloader_options"]["shuffle"] = False

    elif hparams["sorting"] == "random":
        pass

    else:
        raise NotImplementedError(
            "sorting must be random, ascending or descending"
        )

    valid_data = sb.dataio.dataset.DynamicItemDataset.from_csv(
        csv_path=hparams["valid_csv"], replacements={"data_root": data_folder},
    )
    # We also sort the validation data so it is faster to validate
    valid_data = valid_data.filtered_sorted(sort_key="duration")
    test_datasets = {}
    for csv_file in hparams["test_csv"]:
        name = Path(csv_file).stem
        test_datasets[name] = sb.dataio.dataset.DynamicItemDataset.from_csv(
            csv_path=csv_file, replacements={"data_root": data_folder}
        )
        test_datasets[name] = test_datasets[name].filtered_sorted(
            sort_key="duration"
        )

    datasets = [train_data, valid_data] + [i for k, i in test_datasets.items()]


    # 2. Define audio pipeline:
    @sb.utils.data_pipeline.takes("wav")
    @sb.utils.data_pipeline.provides("sig")
    def audio_pipeline(wav):
        info = torchaudio.info(wav)
        sig = sb.dataio.dataio.read_audio(wav)
        resampled = torchaudio.transforms.Resample(
            info.sample_rate, hparams["sample_rate"],
        )(sig)
        return resampled

    sb.dataio.dataset.add_dynamic_item(datasets, audio_pipeline)
    label_encoder = sb.dataio.encoder.CTCTextEncoder()

    # 3. Define text pipeline:
    @sb.utils.data_pipeline.takes("wrd")
    @sb.utils.data_pipeline.provides(
        "wrd", "char_list", "tokens_list", "tokens"
    )
    def text_pipeline(wrd):
        yield wrd
        char_list = list(wrd)
        yield char_list
        tokens_list = label_encoder.encode_sequence(char_list)
        yield tokens_list
        tokens = torch.LongTensor(tokens_list)
        yield tokens

    sb.dataio.dataset.add_dynamic_item(datasets, text_pipeline)
    lab_enc_file = os.path.join(hparams["save_folder"], "label_encoder.txt")
    special_labels = {
        "blank_label": hparams["blank_index"],
        "unk_label": hparams["unk_index"]
    }
    label_encoder.load_or_create(
        path=lab_enc_file,
        from_didatasets=[train_data],
        output_key="char_list",
        special_labels=special_labels,
        sequence_input=True,
    )

    # 4. Set output:
    sb.dataio.dataset.set_output_keys(
        datasets, ["id", "sig", "wrd", "char_list", "tokens"],
    )
    return train_data, valid_data,test_datasets, label_encoder



# Load hyperparameters file with command-line overrides
hparams_file, run_opts, overrides = sb.parse_arguments(["train_semi.yaml"])
with open(hparams_file) as fin:
    hparams = load_hyperpyyaml(fin, overrides)

# If --distributed_launch then
# create ddp_group with the right communication protocol
sb.utils.distributed.ddp_init_group(run_opts)


# Create experiment directory
sb.create_experiment_directory(
    experiment_directory=hparams["output_folder"],
    hyperparams_to_save=hparams_file,
    overrides=overrides,
)

# Due to DDP, we do the preparation ONLY on the main python process
# Defining tokenizer and loading it
# Create the datasets objects as well as tokenization and encoding :-D
label_encoder = sb.dataio.encoder.CTCTextEncoder()

lab_enc_file = os.path.join(hparams["save_folder"], "label_encoder.txt")
special_labels = {
    "blank_label": hparams["blank_index"],
    "unk_label": hparams["unk_index"]
}
label_encoder.load_or_create(
    path=lab_enc_file,
    from_didatasets=[[]],
    output_key="char_list",
    special_labels=special_labels,
    sequence_input=True,
)

from pyctcdecode import build_ctcdecoder
ind2lab = label_encoder.ind2lab
print(ind2lab)
labels = [ind2lab[x] for x in range(len(ind2lab))]
labels = [""] + labels[1:-1] + ["1"] 
# Replace the <blank> token with a blank character, needed for PyCTCdecode
print(labels)
decoder = build_ctcdecoder(
    labels,
    kenlm_model_path=hparams["ngram_lm_path"],  # .arpa or .bin
    alpha=0.5,  # Default by KenLM
    beta=1.0,  # Default by KenLM
)
# Trainer initialization
run_opts["device"] = "cpu"
asr_brain = ASR(
    modules=hparams["modules"],
    hparams=hparams,
    run_opts=run_opts,
    checkpointer=hparams["checkpointer"],
)

# Adding objects to trainer.
asr_brain.tokenizer = label_encoder
asr_brain.checkpointer.recover_if_possible(device="cpu")
asr_brain.modules.eval()
description = """This is a speechbrain-based Automatic Speech Recognition (ASR) model for Tunisian arabic. It outputs Tunisian Arabic transcriptions written in Arabic characters.
This model outputs transcriptions in arabic alphabet only and performs poorly with sentences containing foreign words. However if you do  need code-switching in your transcripts, i.e. foreign outputs in latin alphabet,  you would better use the code switched model, available in another space from the same author. (https://huggingface.co/SalahZa/Code_Switched_Tunisian_Speech_Recognition) 

Run is done on CPU to keep it free in this space. This leads to quite long running times on long sequences. If for your project or research, you want to transcribe long sequences, you would better use the model directly from its page, some instructions for inference on a test set have been provided there. (https://huggingface.co/SalahZa/Tunisian_Automatic_Speech_Recognition). If you need help,  feel free to drop an email here : [email protected] 

Authors : 
* [Salah Zaiem](https://fr.linkedin.com/in/salah-zaiem)
* [Ahmed Amine Ben Aballah](https://www.linkedin.com/in/aabenz/)
* [Ata Kaboudi](https://www.linkedin.com/in/ata-kaboudi-63365b1a8)
* [Amir Kanoun](https://tn.linkedin.com/in/ahmed-amir-kanoun)

More in-depth details and insights are available in a released preprint. Please find the paper [here](https://arxiv.org/abs/2309.11327).
If you use or refer to this model, please cite : 

```
@misc{abdallah2023leveraging,
      title={Leveraging Data Collection and Unsupervised Learning for Code-switched Tunisian Arabic Automatic Speech Recognition}, 
      author={Ahmed Amine Ben Abdallah and Ata Kabboudi and Amir Kanoun and Salah Zaiem},
      year={2023},
      eprint={2309.11327},
      archivePrefix={arXiv},
      primaryClass={eess.AS}
}


"""
title = "Tunisian Speech Recognition"

def treat_wav_file(file_mic,file_upload ,asr=asr_brain, device="cpu") :
    if (file_mic is not None) and (file_upload is not None):
        warn_output = "WARNING: You've uploaded an audio file and used the microphone. The recorded file from the microphone will be used and the uploaded audio will be discarded.\n"
        wav = file_mic
    elif (file_mic is None) and (file_upload is None):
        return "ERROR: You have to either use the microphone or upload an audio file"
    elif file_mic is not None:
        wav = file_mic
    else:
        wav = file_upload
    info = torchaudio.info(wav)
    sr = info.sample_rate
    sig = sb.dataio.dataio.read_audio(wav)
    if len(sig.shape)>1 :
        sig = torch.mean(sig, dim=1)
    sig = torch.unsqueeze(sig, 0)
    tensor_wav = sig.to(device)
    resampled = torchaudio.functional.resample( tensor_wav, sr, 16000)
    sentence = asr.treat_wav(resampled)
    return sentence

gr.Interface(
    title = title,
    description = description,
    fn=treat_wav_file,
    inputs=[gr.Audio(source="microphone", type='filepath', label = "record", optional = True),
            gr.Audio(source="upload", type='filepath', label="filein", optional=True)]
    ,outputs="text").launch()