realzza-meld-asr-hubert-transformer

espnet/realzza-meld-asr-hubert-transformer

realzza-meld-asr-hubert-transformer is a automatic speech recognition model from espnet. Use it when you need speech turned into text. It is set up for espnet. The card lists the license as cc-by-4.0.

- date: Thu Nov 10 09:07:40 EST 2022 - python version: 3.8.6 (default, Dec 17 2020, 16:57:01) [GCC 10.2.0] - espnet version: espnet 202207 - pytorch version: pytorch 1.8.1+cu102 - Git hash: a7bd6522b32ec6472c13f6a2289…

Downloads · 30 days

2% of all-time downloads

All-time downloads

Public

Repo size

1.6 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.pth1.6 GB · 100%

At a glance

Task: Automatic Speech Recognition
Library: espnet
License: cc-by-4.0
Access: Public
Created: Nov 18, 2022
Updated: Sep 19, 2026
SHA: 35564db4

Datasets

meld

Model card

From the Hugging Face model README

Usage

import librosa
from espnet2.bin.asr_inference import Speech2Text

speech2text = Speech2Text.from_pretrained(model_tag="espnet/realzza-meld-asr-hubert-transformer")
# librosa resamples and mixes to one channel, so any file works; 16000 is
# what nearly every espnet recogniser is trained on - check this model's
# config if its audio is not 16 kHz
speech, rate = librosa.load("audio.wav", sr=16000, mono=True)
text, *_ = speech2text(speech)[0]
print(text)

ESPnet2: Meld Recipe

Demo: How to use in ESPnet2

cd espnet
pip install -e .
cd egs2/meld/asr1/
./run.sh

Environments

date: Thu Nov 10 09:07:40 EST 2022
python version: 3.8.6 (default, Dec 17 2020, 16:57:01) [GCC 10.2.0]
espnet version: espnet 202207
pytorch version: pytorch 1.8.1+cu102
Git hash: a7bd6522b32ec6472c13f6a2289dcdff4a846c12
- Commit date: Wed Sep 14 08:34:27 2022 -0400

asr_train_asr_hubert_transformer_adam_specaug_meld_raw_en_bpe850

ASR config: conf/tuning/train_asr_hubert_transformer_adam_specaug_meld.yaml
token_type: bpe
keep_nbest_models: 5

dataset	Snt	Emotion Classification (%)
decoder_asr_asr_model_valid.acc.ave_5best/test	2608	39.22
decoder_asr_asr_model_valid.acc.ave_5best/valid	1104	42.64

ASR results

WER

dataset	Snt	Wrd	Corr	Sub	Del	Ins	Err	S.Err
decoder_asr_asr_model_valid.acc.ave_5best/test	2608	24809	55.5	28.0	16.5	8.4	52.9	96.5
decoder_asr_asr_model_valid.acc.ave_5best/valid	1104	10171	55.3	29.4	15.3	7.0	51.7	96.2

CER

dataset	Snt	Wrd	Corr	Sub	Del	Ins	Err	S.Err
decoder_asr_asr_model_valid.acc.ave_5best/test	2608	120780	71.1	10.7	18.2	10.6	39.5	96.5
decoder_asr_asr_model_valid.acc.ave_5best/valid	1104	49323	71.3	11.1	17.6	9.4	38.1	96.2

TER

dataset	Snt	Wrd	Corr	Sub	Del	Ins	Err	S.Err
decoder_asr_asr_model_valid.acc.ave_5best/test	2608	35287	57.6	21.8	20.5	7.8	50.2	96.5
decoder_asr_asr_model_valid.acc.ave_5best/valid	1104	14430	57.4	23.2	19.4	6.1	48.6	96.2

Task: Automatic Speech Recognition
Library: espnet
License: cc-by-4.0
Languages: en
Created: Nov 18, 2022
Updated: Sep 19, 2026