metadata

language: it
datasets:
  - common_voice
metrics:
  - wer
  - cer
tags:
  - audio
  - automatic-speech-recognition
  - speech
  - xlsr-fine-tuning-week
license: apache-2.0
model-index:
  - name: XLSR Wav2Vec2 Italian by Jonatas Grosman
    results:
      - task:
          name: Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice it
          type: common_voice
          args: it
        metrics:
          - name: Test WER
            type: wer
            value: 9.36
          - name: Test CER
            type: cer
            value: 2.33

Wav2Vec2-Large-XLSR-53-Italian

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Italian using the Common Voice. When using this model, make sure that your speech input is sampled at 16kHz.

The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint

Usage

The model can be used directly (without a language model) as follows:

import torch
import librosa
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor

LANG_ID = "it"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-italian"
SAMPLES = 10

test_dataset = load_dataset("common_voice", LANG_ID, split=f"test[:{SAMPLES}]")

processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
    batch["speech"] = speech_array
    batch["sentence"] = batch["sentence"].upper()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

with torch.no_grad():
    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits

predicted_ids = torch.argmax(logits, dim=-1)
predicted_sentences = processor.batch_decode(predicted_ids)

for i, predicted_sentence in enumerate(predicted_sentences):
    print("-" * 100)
    print("Reference:", test_dataset[i]["sentence"])
    print("Prediction:", predicted_sentence)

Reference	Prediction
POI LEI MORÌ.	POI LEI MORÌ
IL LIBRO HA SUSCITATO MOLTE POLEMICHE A CAUSA DEI SUOI CONTENUTI.	IL LIBRO HA SUSCITATO MOLTE POLEMICHE A CAUSA DEI SUOI CONTENUTI
"FIN DALL'INIZIO LA SEDE EPISCOPALE È STATA IMMEDIATAMENTE SOGGETTA ALLA SANTA SEDE."	FIN DALL'INIZIO LA SEDE EPISCOPALE È STATA IMMEDIATAMENTE SOGGETTA ALLA SANTA SEDE
IL VUOTO ASSOLUTO?	IL VUOTO ASSOLUTO
DOPO ALCUNI ANNI, EGLI DECISE DI TORNARE IN INDIA PER RACCOGLIERE ALTRI INSEGNAMENTI.	DOPO ALCUNI ANNI EGLI DECISE DI TORNARE IN INDIA PER RACCOGLIERE ALTRI INSEGNAMENTI
SALVATION SUE	SALVATION SOO
IN QUESTO MODO, DECIO OTTENNE IL POTERE IMPERIALE.	IN QUESTO MODO DECHO OTTENNE IL POTERE IMPERIALE
SPARTA NOVARA ACQUISISCE IL TITOLO SPORTIVO PER GIOCARE IN PRIMA CATEGORIA.	PARCANOVARACFILISCE IL TITOLO SPORTIVO PER GIOCARE IN PRIMA CATEGORIA
IN SEGUITO, KYGO E SHEAR HANNO PROPOSTO DI CONTINUARE A LAVORARE SULLA CANZONE.	IN SEGUITO KIGO E SHIAR HANNO PROPOSTO DI CONTINUARE A LAVORARE SULLA CANZONE
ALAN CLARKE	ALAN CLARK

Evaluation

The model can be evaluated as follows on the Italian test data of Common Voice.

import torch
import re
import librosa
from datasets import load_dataset, load_metric
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor

LANG_ID = "it"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-italian"
DEVICE = "cuda"

CHARS_TO_IGNORE = [",", "?", "¿", ".", "!", "¡", ";", "；", ":", '""', "%", '"', "�", "ʿ", "·", "჻", "~", "՞",
                   "؟", "،", "।", "॥", "«", "»", "„", "“", "”", "「", "」", "‘", "’", "《", "》", "(", ")", "[", "]",
                   "{", "}", "=", "`", "_", "+", "<", ">", "…", "–", "°", "´", "ʾ", "‹", "›", "©", "®", "—", "→", "。",
                   "、", "﹂", "﹁", "‧", "～", "﹏", "，", "｛", "｝", "（", "）", "［", "］", "【", "】", "‥", "〽",
                   "『", "』", "〝", "〟", "⟨", "⟩", "〜", "：", "！", "？", "♪", "؛", "/", "\\", "º", "−", "^", "ʻ", "ˆ"]

test_dataset = load_dataset("common_voice", LANG_ID, split="test")

wer = load_metric("wer.py") # https://github.com/jonatasgrosman/wav2vec2-sprint/blob/main/wer.py
cer = load_metric("cer.py") # https://github.com/jonatasgrosman/wav2vec2-sprint/blob/main/cer.py

chars_to_ignore_regex = f"[{re.escape(''.join(CHARS_TO_IGNORE))}]"

processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
model.to(DEVICE)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
    with warnings.catch_warnings():
        warnings.simplefilter("ignore")
        speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
    batch["speech"] = speech_array
    batch["sentence"] = re.sub(chars_to_ignore_regex, "", batch["sentence"]).upper()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def evaluate(batch):
    inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

    with torch.no_grad():
        logits = model(inputs.input_values.to(DEVICE), attention_mask=inputs.attention_mask.to(DEVICE)).logits

    pred_ids = torch.argmax(logits, dim=-1)
    batch["pred_strings"] = processor.batch_decode(pred_ids)
    return batch

result = test_dataset.map(evaluate, batched=True, batch_size=8)

predictions = [x.upper() for x in result["pred_strings"]]
references = [x.upper() for x in result["sentence"]]

print(f"WER: {wer.compute(predictions=predictions, references=references, chunk_size=1000) * 100}")
print(f"CER: {cer.compute(predictions=predictions, references=references, chunk_size=1000) * 100}")

Test Result:

In the table below I report the Word Error Rate (WER) and the Character Error Rate (CER) of the model. I ran the evaluation script described above on other models as well (on 2021-04-21). Note that the table below may show different results from those already reported, this may have been caused due to some specificity of the other evaluation scripts used.

Model	WER	CER
jonatasgrosman/wav2vec2-large-xlsr-53-italian	9.36%	2.33%
joorock12/wav2vec2-large-xlsr-italian	12.60%	3.18%
gchhablani/wav2vec2-large-xlsr-it	12.99%	3.11%
facebook/wav2vec2-large-xlsr-53-italian	22.08%	6.36%