SEW by ASAPP Research

The base model pretrained on 16kHz sampled speech audio. When using the model make sure that your speech input is also sampled at 16Khz. Note that this model should be fine-tuned on a downstream task, like Automatic Speech Recognition, Speaker Identification, Intent Classification, Emotion Recognition, etc...

Paper: Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition

Authors: Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi

Abstract This paper is a study of performance-efficiency trade-offs in pre-trained models for automatic speech recognition (ASR). We focus on wav2vec 2.0, and formalize several architecture designs that influence both the model performance and its efficiency. Putting together all our observations, we introduce SEW (Squeezed and Efficient Wav2vec), a pre-trained model architecture with significant improvements along both performance and efficiency dimensions across a variety of training setups. For example, under the 100h-960h semi-supervised setup on LibriSpeech, SEW achieves a 1.9x inference speedup compared to wav2vec 2.0, with a 13.5% relative reduction in word error rate. With a similar inference time, SEW reduces word error rate by 25-50% across different model sizes.

The original model can be found under .


To transcribe audio files the model can be used as a standalone acoustic model as follows:

from transformers import Wav2Vec2Processor, SEWForCTC
from datasets import load_dataset
import soundfile as sf
import torch
# load the model and preprocessor
processor = Wav2Vec2Processor.from_pretrained("asapp/sew-tiny-100k-ft-ls100h")
model = SEWForCTC.from_pretrained("asapp/sew-tiny-100k-ft-ls100h")

# load the dummy dataset with speech samples
ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")
# preprocess
input_values = processor(ds[0]["audio"]["array"], return_tensors="pt").input_values  # Batch size 1

# retrieve logits
logits = model(input_values).logits
# take argmax and decode
predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)


This code snippet shows how to evaluate asapp/sew-tiny-100k-ft-ls100h on LibriSpeech's "clean" and "other" test data.

from datasets import load_dataset
from transformers import SEWForCTC, Wav2Vec2Processor
import torch
from jiwer import wer

librispeech_eval = load_dataset("librispeech_asr", "clean", split="test")

model = SEWForCTC.from_pretrained("asapp/sew-tiny-100k-ft-ls100h").to("cuda")
processor = Wav2Vec2Processor.from_pretrained("asapp/sew-tiny-100k-ft-ls100h")

def map_to_pred(batch):
    input_values = processor(batch["audio"][0]["array"], sampling_rate=16000, 
                             return_tensors="pt", padding="longest").input_values
    with torch.no_grad():
        logits = model("cuda")).logits

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

result =, batched=True, batch_size=1, remove_columns=["audio"])

print("WER:", wer(result["text"], result["transcription"]))

Result (WER):

"clean" "other"
10.61 23.74
Downloads last month
Hosted inference API
This model can be loaded on the Inference API on-demand.