Upload folder using huggingface_hub

.gitattributes

@@ -79,3 +79,83 @@ examples/2SQA/Spoken-Squad-Test/sample_1.wav filter=lfs diff=lfs merge=lfs -text
 examples/2SQA/Spoken-Squad-Test/sample_2.wav filter=lfs diff=lfs merge=lfs -text
 examples/SQA/DREAM-TTS-MCQ-Test/sample_1.wav filter=lfs diff=lfs merge=lfs -text
 examples/**/*.arrow filter=lfs diff=lfs merge=lfs -text
+examples/cna_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_17_id_asr/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_17_id_asr/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_17_ta_asr/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_17_ta_asr/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_17_th_asr/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_17_vi_asr/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_17_vi_asr/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_zh_asr/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/commonvoice_zh_asr/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/fleurs_tamil_ta_30_asr/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/fleurs_tamil_ta_30_asr/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/gigaspeech2_id_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/gigaspeech2_th_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/gigaspeech2_th_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/gigaspeech2_vi_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/gigaspeech2_vi_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/idpc_short_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/idpc_short_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_ar_dialogue/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_ar_dialogue/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_ar_sentence/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_ar_sentence/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part1_asr_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part1_asr_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part2_asr_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part2_asr_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part3_30s_asr_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part3_30s_asr_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part3_30s_ds_human_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part3_30s_ds_human_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part3_30s_sqa_human_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part3_30s_sqa_human_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part4_30s_asr_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part4_30s_asr_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part4_30s_ds_human_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part4_30s_ds_human_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part4_30s_sqa_human_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part4_30s_sqa_human_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part5_30s_asr_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part5_30s_asr_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part5_30s_ds_human_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part5_30s_ds_human_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part5_30s_sqa_human_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part5_30s_sqa_human_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part6_30s_asr_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part6_30s_asr_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part6_30s_ds_human_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part6_30s_ds_human_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part6_30s_sqa_human_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/imda_part6_30s_sqa_human_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/lotus_thai_th_30_asr/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/lotus_thai_th_30_asr/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/mediacorp_short_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/mediacorp_short_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/mediacorp_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/mediacorp_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/parliament_short_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/parliament_short_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/seame_dev_sge/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ukusnews_short_test/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ukusnews_short_test/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch1/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch1/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch2/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch2/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch3_chinese/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch3_chinese/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch3_malay/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch3_malay/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch3_tamil/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch3_tamil/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch3_tamil_v2/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_asr_batch3_tamil_v2/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_pqa_batch1/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_pqa_batch1/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_sds_batch1/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_sds_batch1/example_1.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_sqa_batch1/example_0.wav filter=lfs diff=lfs merge=lfs -text
+examples/ytb_sqa_batch1/example_1.wav filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -1,13 +0,0 @@
-auto_evals/
-venv/
-__pycache__/
-.env
-.ipynb_checkpoints
-*ipynb
-.vscode/
-
-eval-queue/
-eval-results/
-eval-queue-bk/
-eval-results-bk/
-logs/

.pre-commit-config.yaml

@@ -1,53 +0,0 @@
-# Copyright (c) 2022, NVIDIA CORPORATION.  All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-default_language_version:
-  python: python3
-
-ci:
-  autofix_prs: true
-  autoupdate_commit_msg: '[pre-commit.ci] pre-commit suggestions'
-  autoupdate_schedule: quarterly
-
-repos:
-  - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.3.0
-    hooks:
-      - id: check-yaml
-      - id: check-case-conflict
-      - id: detect-private-key
-      - id: check-added-large-files
-        args: ['--maxkb=1000']
-      - id: requirements-txt-fixer
-      - id: end-of-file-fixer
-      - id: trailing-whitespace
-
-  - repo: https://github.com/PyCQA/isort
-    rev: 5.12.0
-    hooks:
-      - id: isort
-        name: Format imports
-
-  - repo: https://github.com/psf/black
-    rev: 22.12.0
-    hooks:
-      - id: black
-        name: Format code
-        additional_dependencies: ['click==8.0.2']
-
-  - repo: https://github.com/charliermarsh/ruff-pre-commit
-    # Ruff version.
-    rev: 'v0.0.267'
-    hooks:
-      - id: ruff

Makefile

@@ -1,13 +0,0 @@
-.PHONY: style format
-
-
-style:
-	python -m black --line-length 119 .
-	python -m isort .
-	ruff check --fix .
-
-
-quality:
-	python -m black --check --line-length 119 .
-	python -m isort --check-only .
-	ruff check .

README.md

@@ -1,45 +1,9 @@
 ---
-title: Leaderboard / AudioBench
+title: AudioBench Leaderboard
 emoji: 🥇
 colorFrom: green
 colorTo: indigo
-sdk: streamlit
-app_file: app.py
+sdk: static
 pinned: false
-sdk_version: 1.37.1
 license: apache-2.0
 ---
-
-# Start the configuration
-
-Most of the variables to change for a default leaderboard are in `src/env.py` (replace the path for your leaderboard) and `src/about.py` (for tasks).
-
-Results files should have the following format and be stored as json files:
-```json
-{
-    "config": {
-        "model_dtype": "torch.float16", # or torch.bfloat16 or 8bit or 4bit
-        "model_name": "path of the model on the hub: org/model",
-        "model_sha": "revision on the hub",
-    },
-    "results": {
-        "task_name": {
-            "metric_name": score,
-        },
-        "task_name2": {
-            "metric_name": score,
-        }
-    }
-}
-```
-
-Request files are created automatically by this tool.
-
-If you encounter problem on the space, don't hesitate to restart it to remove the create eval-queue, eval-queue-bk, eval-results and eval-results-bk created folder.
-
-# Code logic for more complex edits
-
-You'll find 
-- the main table' columns names and properties in `src/display/utils.py`
-- the logic to read all results and request files, then convert them in dataframe lines, in `src/leaderboard/read_evals.py`, and `src/populate.py`
-- teh logic to allow or filter submissions in `src/submission/submit.py` and `src/submission/check_validity.py`

app.py

@@ -1,112 +0,0 @@
-import streamlit as st
-import streamlit_antd_components as sac
-
-from app.pages import *
-
-
-# Set page configuration
-st.set_page_config(
-    page_title="AudioBench Leaderboard",
-    page_icon=":chart_with_upwards_trend:",
-    layout="wide",
-)
-
-
-
-# Dictionary mapping menu items to their corresponding functions
-pages = {
-    'Dashboard'          : dashboard,
-    'ASR-English'        : asr_english,
-    'ASR-Mandarin'       : asr_mandarin,
-    'ASR-Singlish'       : asr_singlish,
-    'ASR-Malay'          : asr_malay,
-    'ASR-Tamil'          : asr_tamil,
-    'ASR-Indonesian'     : asr_indonesian,
-    'ASR-Thai'           : asr_thai,
-    'ASR-Vietnamese'     : asr_vietnamese,
-    'ASR-Other'        : asr_private,
-    'Speech Translation' : speech_translation,
-    'SQA-English'        : speech_question_answering_english,
-    'SQA-Singlish'       : speech_question_answering_singlish,
-    'SDS-Singlish'       : spoken_dialogue_summarization_singlish,
-    'Speech Instruction' : speech_instruction,
-    'Audio Captioning'   : audio_captioning,
-    'Audio-Scene QA'     : audio_scene_question_answering,
-    'Accent Recognition' : accent_recognition,
-    'Gender Recognition' : gender_recognition,
-    'Emotion Recognition': emotion_recognition,
-    'Music Understanding': music_understanding,
-
-    '* Under Development *': under_development,
-}
-
-# Initialize session state for menu selection
-if 'selected_menu' not in st.session_state:
-    st.session_state.selected_menu = 'Introduction'
-
-# Define the menu items
-menu_items = [
-    sac.MenuItem(label='Dashboard', icon='house'),
-
-    sac.MenuItem(label='Automatic Speech Recognition', icon='mic',
-                 children = [
-                     sac.MenuItem(label='ASR-English', icon='mic'),
-                     sac.MenuItem(label='ASR-Singlish', icon='mic'),
-                     sac.MenuItem(label='ASR-Mandarin', icon='mic'),
-                     sac.MenuItem(label='ASR-Malay', icon='mic'),
-                     sac.MenuItem(label='ASR-Tamil', icon='mic'),
-                     sac.MenuItem(label='ASR-Indonesian', icon='mic'),
-                     sac.MenuItem(label='ASR-Thai', icon='mic'),
-                     sac.MenuItem(label='ASR-Vietnamese', icon='mic'),
-                     sac.MenuItem(label='ASR-Other', icon='mic'),
-                 ]
-                 ),
-
-    sac.MenuItem(label='Speech Translation', icon='translate'
-                 ),
-
-    sac.MenuItem(label='Spoken Question Answering', icon='question-circle',
-                 children = [
-                     sac.MenuItem(label='SQA-English', icon='mic'),
-                     sac.MenuItem(label='SQA-Singlish', icon='mic'),
-                 ]
-                 ),
-
-    sac.MenuItem(label='Spoken Dialogue Summarization', icon='question-circle',
-                 children = [
-                     sac.MenuItem(label='SDS-Singlish', icon='mic'),
-                 ]
-                 ),
-
-    sac.MenuItem(label='Speech Instruction', icon='mic-fill'),
-
-    sac.MenuItem(label='Audio Captioning', icon='volume-down'),
-
-    sac.MenuItem(label='Audio-Scene QA', icon='question-diamond-fill'),
-    
-    sac.MenuItem(label='Accent Recognition', icon='person-badge-fill'),
-    
-    sac.MenuItem(label='Gender Recognition', icon='gender-ambiguous'),
-
-    sac.MenuItem(label='Emotion Recognition', icon='emoji-smile-fill'),
-
-    sac.MenuItem(label='Music Understanding', icon='music-note-list'),
-
-    sac.MenuItem(label='* Under Development *', icon='lock'),
-
-]
-
-# Render the menu in the sidebar
-with st.sidebar:
-    selected = sac.menu(menu_items,
-                        size='sm', 
-                        open_all=False,
-                        )
-
-# Update session state based on selection
-if selected:
-    st.session_state.selected_menu = selected
-
-# Display the selected page's content
-page = pages[st.session_state.selected_menu]
-page()

app/content.py

@@ -1,431 +0,0 @@
-asr_english_datasets = [
-    'LibriSpeech-Clean', 
-    'LibriSpeech-Other', 
-    'CommonVoice-15-EN', 
-    'Peoples-Speech', 
-    'GigaSpeech-1', 
-    'Earnings-21', 
-    'Earnings-22', 
-    'TED-LIUM-3', 
-    'TED-LIUM-3-LongForm', 
-]
-
-
-asr_singlish_datasets = [
-    'MNSC-PART1-ASR', 
-    'MNSC-PART2-ASR',
-    'MNSC-PART3-ASR',
-    'MNSC-PART4-ASR',
-    'MNSC-PART5-ASR',
-    'MNSC-PART6-ASR',
-]
-
-
-asr_mandarin_datasets = [
-    'AISHELL-ASR-ZH', 
-    'CommonVoice-ZH',
-    'YouTube ASR: Chinese with English Prompt',
-]
-
-
-asr_malay_datasets = [
-    'YouTube ASR: Malay with English Prompt'
-]
-
-
-asr_tamil_datasets = [
-    'CommonVoice-17-Tamil',
-    'Fleurs-Tamil',
-    'YouTube ASR: Tamil with English Prompt'
-]
-
-
-asr_indonesian_datasets = [
-    'CommonVoice-17-Indonesian', 
-    'GigaSpeech-2-Indonesain',
-]
-
-
-asr_thai_datasets = [
-    'GigaSpeech-2-Thai',
-    'Lotus-Thai'
-]
-
-
-asr_vietnamese_datasets = [
-    'CommonVoice-17-Vietnamese',
-    'GigaSpeech-2-Vietnamese'
-]
-
-
-asr_private_datasets =  [
-    'CNA',
-    'IDPC',
-    'Parliament',
-    'UKUS-News',
-    'Mediacorp',
-    'IDPC-Short',
-    'Parliament-Short',
-    'UKUS-News-Short',
-    'Mediacorp-Short',
-    'YouTube ASR: English Singapore Content',
-    'YouTube ASR: English with Strong Emotion',
-]
-
-
-speech_translation_datasets = [
-    'CoVoST2-EN-ID', 
-    'CoVoST2-EN-ZH',
-    'CoVoST2-EN-TA', 
-    'CoVoST2-ID-EN', 
-    'CoVoST2-ZH-EN', 
-    'CoVoST2-TA-EN'
-]
-
-
-speech_qa_english_datasets = [
-    'CN-College-Listen-MCQ',
-    'DREAM-TTS-MCQ',
-    'SLUE-P2-SQA5', 
-    'Public-SG-Speech-QA', 
-    'Spoken-SQuAD',
-    'MMAU-mini'
-]
-
-
-speech_qa_singlish_datasets = [
-    'MNSC-PART3-SQA', 
-    'MNSC-PART4-SQA',
-    'MNSC-PART5-SQA',
-    'MNSC-PART6-SQA',
-]
-
-
-sds_datasets = [
-    'MNSC-PART3-SDS', 
-    'MNSC-PART4-SDS',
-    'MNSC-PART5-SDS',
-    'MNSC-PART6-SDS',
-]
-
-
-si_datasets = [
-    'OpenHermes-Audio', 
-    'ALPACA-Audio',
-]
-
-
-ac_datasets = [
-    'WavCaps', 
-    'AudioCaps',
-]
-
-
-asqa_datasets = [
-    'Clotho-AQA', 
-    'WavCaps-QA', 
-    'AudioCaps-QA'
-]
-
-
-er_datasets = [
-    'IEMOCAP-Emotion', 
-    'MELD-Sentiment', 
-    'MELD-Emotion',
-]
-
-
-ar_datasets = [
-    'VoxCeleb-Accent',
-    'MNSC-AR-Sentence',
-    'MNSC-AR-Dialogue',
-]
-
-
-gr_datasets = [
-    'VoxCeleb-Gender', 
-    'IEMOCAP-Gender'
-]
-
-
-music_datasets = ['MuChoMusic']
-
-
-wer_development_datasets = [
-    'YouTube ASR: Malay with Malay Prompt',
-    'YouTube ASR: Chinese with Chinese Prompt',
-    'SEAME-Dev-Mandarin',
-    'SEAME-Dev-Singlish',
-]
-
-
-non_wer_development_datasets = [
-    'YouTube SQA: English with Singapore Content',
-    'YouTube SDS: English with Singapore Content',
-    'YouTube PQA: English with Singapore Content',
-]
-
-
-wer_displayname2datasetname = {
-    'LibriSpeech-Clean'    : 'librispeech_test_clean',
-    'LibriSpeech-Other'    : 'librispeech_test_other',
-    'CommonVoice-15-EN'    : 'common_voice_15_en_test',
-    'Peoples-Speech'       : 'peoples_speech_test',
-    'GigaSpeech-1'         : 'gigaspeech_test',
-    'Earnings-21'          : 'earnings21_test',
-    'Earnings-22'          : 'earnings22_test',
-    'TED-LIUM-3'           : 'tedlium3_test',
-    'TED-LIUM-3-LongForm'  : 'tedlium3_long_form_test',
-
-    'MNSC-PART1-ASR'       : 'imda_part1_asr_test',
-    'MNSC-PART2-ASR'       : 'imda_part2_asr_test',
-    'MNSC-PART3-ASR'       : 'imda_part3_30s_asr_test',
-    'MNSC-PART4-ASR'       : 'imda_part4_30s_asr_test',
-    'MNSC-PART5-ASR'       : 'imda_part5_30s_asr_test',
-    'MNSC-PART6-ASR'       : 'imda_part6_30s_asr_test',
-
-    'AISHELL-ASR-ZH'       : 'aishell_asr_zh_test',
-    'CommonVoice-ZH'       : 'commonvoice_zh_asr',
-
-    'CommonVoice-17-Indonesian'    : 'commonvoice_17_id_asr',
-    'CommonVoice-17-Tamil'         : 'commonvoice_17_ta_asr',
-    'CommonVoice-17-Thai'          : 'commonvoice_17_th_asr',
-    'CommonVoice-17-Vietnamese'    : 'commonvoice_17_vi_asr',
-    'GigaSpeech-2-Indonesain'      : 'gigaspeech2_id_test',
-    'GigaSpeech-2-Thai'            : 'gigaspeech2_th_test',
-    'GigaSpeech-2-Vietnamese'      : 'gigaspeech2_vi_test',
-    'Fleurs-Tamil'                 : 'fleurs_tamil_ta_30_asr',
-    'Lotus-Thai'                   : 'lotus_thai_th_30_asr',
-
-    'CNA'             : 'cna_test',
-    'IDPC'            : 'idpc_test',
-    'Parliament'      : 'parliament_test',
-    'UKUS-News'       : 'ukusnews_test',
-    'Mediacorp'       : 'mediacorp_test',
-    'IDPC-Short'      : 'idpc_short_test',
-    'Parliament-Short': 'parliament_short_test',
-    'UKUS-News-Short' : 'ukusnews_short_test',
-    'Mediacorp-Short' : 'mediacorp_short_test',
-    
-    'YouTube ASR: English Singapore Content': 'ytb_asr_batch1',
-    'YouTube ASR: English with Strong Emotion': 'ytb_asr_batch2',
-    'YouTube ASR: Malay with English Prompt': 'ytb_asr_batch3_malay',
-    'YouTube ASR: Chinese with English Prompt': 'ytb_asr_batch3_chinese',
-    'YouTube ASR: Tamil with English Prompt': 'ytb_asr_batch3_tamil',
-    'YouTube ASR: Tamil with English Prompt V2': 'ytb_asr_batch3_tamil_v2',
-
-    'YouTube ASR: Malay with Malay Prompt': 'ytb_asr_batch3_ms_ms_prompt',
-    'YouTube ASR: Chinese with Chinese Prompt': 'ytb_asr_batch3_zh_zh_prompt',
-    
-    'SEAME-Dev-Mandarin'   : 'seame_dev_man',
-    'SEAME-Dev-Singlish'   : 'seame_dev_sge',
-}
-
-
-non_wer_displayname2datasetname = {
-    'CoVoST2-EN-ID'        : 'covost2_en_id_test',
-    'CoVoST2-EN-ZH'        : 'covost2_en_zh_test',
-    'CoVoST2-EN-TA'        : 'covost2_en_ta_test',
-    'CoVoST2-ID-EN'        : 'covost2_id_en_test',
-    'CoVoST2-ZH-EN'        : 'covost2_zh_en_test',
-    'CoVoST2-TA-EN'        : 'covost2_ta_en_test',
-
-    'CN-College-Listen-MCQ': 'cn_college_listen_mcq_test',
-    'DREAM-TTS-MCQ'        : 'dream_tts_mcq_test',
-    'SLUE-P2-SQA5'         : 'slue_p2_sqa5_test',
-    'Public-SG-Speech-QA'  : 'public_sg_speech_qa_test',
-    'Spoken-SQuAD'         : 'spoken_squad_test',
-    'MMAU-mini'            : 'mmau_mini',
-
-    'MNSC-PART3-SQA'       : 'imda_part3_30s_sqa_human_test',
-    'MNSC-PART4-SQA'       : 'imda_part4_30s_sqa_human_test',
-    'MNSC-PART5-SQA'       : 'imda_part5_30s_sqa_human_test',
-    'MNSC-PART6-SQA'       : 'imda_part6_30s_sqa_human_test',
-
-    'MNSC-PART3-SDS'       : 'imda_part3_30s_ds_human_test',
-    'MNSC-PART4-SDS'       : 'imda_part4_30s_ds_human_test',
-    'MNSC-PART5-SDS'       : 'imda_part5_30s_ds_human_test',
-    'MNSC-PART6-SDS'       : 'imda_part6_30s_ds_human_test',
-
-    'OpenHermes-Audio'     : 'openhermes_audio_test',
-    'ALPACA-Audio'         : 'alpaca_audio_test',
-
-    'WavCaps'              : 'wavcaps_test',
-    'AudioCaps'            : 'audiocaps_test',
-
-    'Clotho-AQA'           : 'clotho_aqa_test',
-    'WavCaps-QA'           : 'wavcaps_qa_test',
-    'AudioCaps-QA'         : 'audiocaps_qa_test',
-
-    'IEMOCAP-Emotion'      : 'iemocap_emotion_test',
-    'MELD-Sentiment'       : 'meld_sentiment_test',
-    'MELD-Emotion'         : 'meld_emotion_test',
-
-    'VoxCeleb-Accent'      : 'voxceleb_accent_test',
-    'MNSC-AR-Sentence'     : 'imda_ar_sentence',
-    'MNSC-AR-Dialogue'     : 'imda_ar_dialogue',
-
-    'VoxCeleb-Gender'      : 'voxceleb_gender_test',
-    'IEMOCAP-Gender'       : 'iemocap_gender_test',
-
-    'MuChoMusic'           : 'muchomusic_test',
-
-    'YouTube SQA: English with Singapore Content': 'ytb_sqa_batch1',
-    'YouTube SDS: English with Singapore Content': 'ytb_sds_batch1',
-    'YouTube PQA: English with Singapore Content': 'ytb_pqa_batch1',
-    
-    'YouTube SQA: Malay': 'ytb_sqa_batch3_malay',
-    'YouTube SQA: Chinese': 'ytb_sqa_batch3_chinese',
-    'YouTube SQA: Tamil': 'ytb_sqa_batch3_tamil',
-    
-    'YouTube SDS: Malay': 'ytb_sds_batch3_malay',
-    'YouTube SDS: Chinese': 'ytb_sds_batch3_chinese',
-    'YouTube SDS: Tamil': 'ytb_sds_batch3_tamil',
-}
-
-
-displayname2datasetname = {}
-displayname2datasetname.update(wer_displayname2datasetname)
-displayname2datasetname.update(non_wer_displayname2datasetname)
-
-
-datasetname2diaplayname = {datasetname: displayname for displayname, datasetname in displayname2datasetname.items()}
-
-
-dataset_diaplay_information = {
-    'LibriSpeech-Clean'    : 'A clean, high-quality testset of the LibriSpeech dataset, used for ASR testing.',
-    'LibriSpeech-Other'    : 'A more challenging, noisier testset of the LibriSpeech dataset for ASR testing.',
-    'CommonVoice-15-EN'    : 'Test set from the Common Voice project, which is a crowd-sourced, multilingual speech dataset.',
-    'Peoples-Speech'       : 'A large-scale, open-source speech recognition dataset, with diverse accents and domains.',
-    'GigaSpeech-1'         : 'A large-scale ASR dataset with diverse audio sources like podcasts, interviews, etc.',
-    'Earnings-21'          : 'ASR test dataset focused on earnings calls from 2021, with professional speech and financial jargon.',
-    'Earnings-22'          : 'Similar to Earnings21, but covering earnings calls from 2022.',
-    'TED-LIUM-3'           : 'A test set derived from TED talks, covering diverse speakers and topics.',
-    'TED-LIUM-3-LongForm'  : 'A longer version of the TED-LIUM dataset, containing extended audio samples. This poses challenges to existing fusion methods in handling long audios. However, it provides benchmark for future development.',
-    'AISHELL-ASR-ZH'       : 'ASR test dataset for Mandarin Chinese, based on the Aishell dataset.',
-    'CoVoST2-EN-ID'        : 'CoVoST 2 dataset for speech translation from English to Indonesian.',
-    'CoVoST2-EN-ZH'        : 'CoVoST 2 dataset for speech translation from English to Chinese.',
-    'CoVoST2-EN-TA'        : 'CoVoST 2 dataset for speech translation from English to Tamil.',
-    'CoVoST2-ID-EN'        : 'CoVoST 2 dataset for speech translation from Indonesian to English.',
-    'CoVoST2-ZH-EN'        : 'CoVoST 2 dataset for speech translation from Chinese to English.',
-    'CoVoST2-TA-EN'        : 'CoVoST 2 dataset for speech translation from Tamil to English.',
-    'CN-College-Listen-MCQ': 'Chinese College English Listening Test, with multiple-choice questions.',
-    'DREAM-TTS-MCQ'        : 'DREAM dataset for spoken question-answering, derived from textual data and synthesized speech.',
-    'SLUE-P2-SQA5'         : 'Spoken Language Understanding Evaluation (SLUE) dataset, part 2, focused on QA tasks.',
-    'Public-SG-Speech-QA'  : 'Public dataset for speech-based question answering, gathered from Singapore.',
-    'Spoken-SQuAD'         : 'Spoken SQuAD dataset, based on the textual SQuAD dataset, converted into audio.',
-    'OpenHermes-Audio'     : 'Test set for spoken instructions. Synthesized from the OpenHermes dataset.',
-    'ALPACA-Audio'         : 'Spoken version of the ALPACA dataset, used for evaluating instruction following in audio.',
-    'WavCaps'              : 'WavCaps is a dataset for testing audio captioning, where models generate textual descriptions of audio clips.',
-    'AudioCaps'            : 'AudioCaps dataset, used for generating captions from general audio events.',
-    'Clotho-AQA'           : 'Clotho dataset adapted for audio-based question answering, containing audio clips and questions.',
-    'WavCaps-QA'           : 'Question-answering test dataset derived from WavCaps, focusing on audio content.',
-    'AudioCaps-QA'         : 'AudioCaps adapted for question-answering tasks, using audio events as input for Q&A.',
-    'VoxCeleb-Accent'      : 'Test dataset for accent recognition, based on VoxCeleb, a large speaker identification dataset.',
-    'MNSC-AR-Sentence'     : 'Accent recognition based on the IMDA NSC dataset, focusing on sentence-level accents.',
-    'MNSC-AR-Dialogue'     : 'Accent recognition based on the IMDA NSC dataset, focusing on dialogue-level accents.',
-
-    'VoxCeleb-Gender': 'Test dataset for gender classification, also derived from VoxCeleb.',
-    'IEMOCAP-Gender' : 'Gender classification based on the IEMOCAP dataset.',
-    'IEMOCAP-Emotion': 'Emotion recognition test data from the IEMOCAP dataset, focusing on identifying emotions in speech.',
-    'MELD-Sentiment' : 'Sentiment recognition from speech using the MELD dataset, classifying positive, negative, or neutral sentiments.',
-    'MELD-Emotion'   : 'Emotion classification in speech using MELD, detecting specific emotions like happiness, anger, etc.',
-    'MuChoMusic'     : 'Test dataset for music understanding, from paper: MuChoMusic: Evaluating Music Understanding in Multimodal Audio-Language Models.',
-    'MNSC-PART1-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 1.',
-    'MNSC-PART2-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 2.',
-    'MNSC-PART3-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 3.',
-    'MNSC-PART4-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 4.',
-    'MNSC-PART5-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 5.',
-    'MNSC-PART6-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 6.',
-    'MNSC-PART3-SQA' : 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 3.',
-    'MNSC-PART4-SQA' : 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 4.',
-    'MNSC-PART5-SQA' : 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 5.',
-    'MNSC-PART6-SQA' : 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 6.',
-    'MNSC-PART3-SDS' : 'Multitak National Speech Corpus (MNSC) dataset, dialogue summarization task, Part 3.',
-    'MNSC-PART4-SDS' : 'Multitak National Speech Corpus (MNSC) dataset, dialogue summarization task, Part 4.',
-    'MNSC-PART5-SDS' : 'Multitak National Speech Corpus (MNSC) dataset, dialogue summarization task, Part 5.',
-    'MNSC-PART6-SDS' : 'Multitak National Speech Corpus (MNSC) dataset, dialogue summarization task, Part 6.',
-
-    'CNA'             : 'Under Development',
-    'IDPC'            : 'Under Development',
-    'Parliament'      : 'Under Development',
-    'UKUS-News'       : 'Under Development',
-    'Mediacorp'       : 'Under Development',
-    'IDPC-Short'      : 'Under Development',
-    'Parliament-Short': 'Under Development',
-    'UKUS-News-Short' : 'Under Development',
-    'Mediacorp-Short' : 'Under Development',
-
-    'CommonVoice-ZH'               : 'Under Development',
-    'CommonVoice-17-Indonesian'    : 'Under Development',
-    'CommonVoice-17-Tamil'         : 'Under Development',
-    'CommonVoice-17-Thai'          : 'Under Development',
-    'CommonVoice-17-Vietnamese'    : 'Under Development',
-    'GigaSpeech-2-Indonesain'      : 'Under Development',
-    'GigaSpeech-2-Thai'            : 'Under Development',
-    'GigaSpeech-2-Vietnamese'      : 'Under Development',
-    'Fleurs-Tamil'                 : 'Under Development',
-    'Lotus-Thai'                   : 'Under Development',
-    'MMAU-mini'                    : 'Under Development',
-
-    
-    'YouTube ASR: English Singapore Content'  : 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains English and Singlish audio clips, featuring Singapore-related content. <br> It includes approximately 2.5 hours of audio, with individual clips ranging from 2 seconds to 30 seconds in length.',
-    
-    'YouTube ASR: English with Strong Emotion'  : 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains English, Singlish and some unknown languages audio clips, featuring speech with strong emotional expression. <br> It includes approximately 3.9 hours of audio, with each clip lasting 30 seconds.',
-    
-    'YouTube ASR: Malay with English Prompt': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset mainly contains Malay and some Malay-English codeswitch audio clips, featuring with English prompts. <br> It includes approximately 2.55 hours of audio, with indicidual clips ranging form 30 seconds to 95 seconds in length.',
-    
-    'YouTube ASR: Malay with Malay Prompt': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset use the same audio from <i>YouTube ASR: Malay English Prompt</i>, except featuring with Malay prompts. <br> It includes approximately 2.55 hours of audio, with indicidual clips ranging form 30 seconds to 95 seconds in length.',
-    
-    'YouTube ASR: Chinese with English Prompt': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains Chinese and some Chinese-English codeswitch audio clips, featuring with English prompts. <br> It includes approximately 3.32 hours of audio, with individual clips ranging from 17 seconds to 1966 seconds in length.',
-    
-    'YouTube ASR: Chinese with Chinese Prompt': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains Chinese and some Chinese-English codeswitch audio clips, featuring with Chinese prompts. <br> It includes approximately 3.32 hours of audio, with individual clips ranging from 17 seconds to 1966 seconds in length.',
-    
-    'YouTube ASR: Tamil with Tamil Prompt': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains Tamil and some Tamil-English codeswitch audio clips, featuring with Tamil prompts. <br> It includes approximately 2.44 hours of audio, with individual clips ranging from 30 seconds to 324 seconds in length.',
-    
-    'YouTube ASR: Tamil with English Prompt': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains Tamil and some Tamil-English codeswitch audio clips, featuring with English prompts. <br> It includes approximately 2.44 hours of audio, with individual clips ranging from 30 seconds to 324 seconds in length.',
-    
-    'YouTube ASR: Tamil with English Prompt V2': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains Tamil and some Tamil-English codeswitch audio clips, featuring with English prompts. <br> It includes approximately 2.44 hours of audio, with individual clips ranging from 30 seconds to 324 seconds in length.',
-
-    'YouTube ASR Translation: Malay2English': 'YouTube Evaluation Dataset for ASR Task: <br> The audio of dataset is same as <i>YouTube ASR: Malay<i>',
-    
-    # 'YouTube ASR Translation: Chinese2English': 'YouTube Evaluation Dataset for ASR Task: <br> The audio of dataset is same as <i>YouTube ASR: Chinese<i>',
-    
-    # 'YouTube ASR Translation: Tamil2English': 'YouTube Evaluation Dataset for ASR Task: <br> The audio of dataset is same as <i>YouTube ASR: Tamil<i>',
-    
-    
-    
-    'SEAME-Dev-Mandarin'   : 'Under Development',
-    'SEAME-Dev-Singlish'   : 'Under Development',
-
-    'YouTube SQA: English with Singapore Content': 'YouTube Evaluation Dataset for Speech-QA Task: <br> This dataset contains English and Singlish audio clips, featuring Singapore-related content. <br> It includes approximately 7.6 hours of audio, with individual clips ranging from 8 seconds to 32 seconds in length.',
-    
-    'YouTube SQA: Malay': 'YouTube Evaluation Dataset for Speech-QA Task: <br> The auido of this dataset is same as <i>YouTube ASR: Malay<i>, it contains Malay and some Malay-English codeswitch audio clips, featuring with English prompts. <br> It includes approximately 2.55 hours of audio, with indicidual clips ranging form 30 seconds to 95 seconds in length.',
-    
-    'YouTube SQA: Chinese': 'YouTube Evaluation Dataset for Speech-QA Task: <br> The auido of this dataset is same as <i>YouTube ASR: Chinese<i>',
-    
-    'YouTube SQA: Tamil': 'YouTube Evaluation Dataset for Speech-QA Task: <br> The auido of this dataset is same as <i>YouTube ASR: Tamil<i>',
-    
-    'YouTube SDS: English with Singapore Content': 'YouTube Evaluation Dataset for Summary Task: <br> This dataset contains English and Singlish audio clips, featuring Singapore-related content. <br> It includes approximately 5.4 hours of audio, with individual clips ranging from 8 seconds to 32 seconds in length.',
-    
-    'YouTube SDS: Malay': 'YouTube Evaluation Dataset for Speech-QA Task: <br> The auido of this dataset is same as <i>YouTube ASR: Malay<i>, it contains Malay and some Malay-English codeswitch audio clips, featuring with English prompts. <br> It includes approximately 2.55 hours of audio, with indicidual clips ranging form 30 seconds to 95 seconds in length.',
-    
-    'YouTube SDS: Chinese': 'YouTube Evaluation Dataset for Speech-QA Task: <br> The auido of this dataset is same as <i>YouTube ASR: Chinese<i>',
-    
-    'YouTube SDS: Tamil': 'YouTube Evaluation Dataset for Speech-QA Task: <br> The auido of this dataset is same as <i>YouTube ASR: Tamil<i>',
-    
-    
-    'YouTube PQA: English with Singapore Content': 'YouTube Evaluation Dataset for Paralinguistics QA Task: <br> This dataset contains English and Singlish audio clips, featuring Singapore-related content. <br> It includes approximately 41.4 hours of audio, with individual clips ranging from 41 seconds to 83 seconds in length.',
-    
-    
-                }
-
-
-metrics_info = {
-    'wer'                    : 'Word Error Rate (WER) - The Lower, the better.',
-    'llama3_70b_judge_binary': 'Model-as-a-Judge Peformance. Using LLAMA-3-70B. Scale from 0-100. The higher, the better.',
-    'llama3_70b_judge'       : 'Model-as-a-Judge Peformance. Using LLAMA-3-70B. Scale from 0-100. The higher, the better.',
-    'meteor'                 : 'METEOR Score. The higher, the better.',
-    'bleu'                   : 'BLEU Score. The higher, the better.',
-}
-

app/draw_diagram.py

@@ -1,193 +0,0 @@
-import streamlit as st
-import pandas as pd
-import numpy as np
-from streamlit_echarts import st_echarts
-from app.show_examples import *
-from app.content import *
-
-import pandas as pd
-
-from app.content import wer_displayname2datasetname
-from model_information import get_dataframe
-info_df = get_dataframe()
-
-
-def draw(folder_name, category_name, displayname, metrics, cus_sort=True):
-    
-    folder = f"./results_organized/{metrics}/"
-
-    # Load the results from CSV
-    data_path = f'{folder}/{category_name.lower()}.csv'
-    chart_data = pd.read_csv(data_path).round(3)
-    
-    dataset_name = displayname2datasetname[displayname]
-    chart_data = chart_data[['Model', dataset_name]]
-
-    # Rename to proper display name
-    chart_data = chart_data.rename(columns=datasetname2diaplayname)
-
-    st.markdown("""
-                <style>
-                .stMultiSelect [data-baseweb=select] span {
-                    max-width: 800px;
-                    font-size: 0.9rem;
-                    background-color: #3C6478 !important; /* Background color for selected items */
-                    color: white; /* Change text color */
-                    back
-                }
-                </style>
-                """, unsafe_allow_html=True)
-    
-    # remap model names
-    display_model_names = {key.strip() :val.strip() for key, val in zip(info_df['Original Name'], info_df['Proper Display Name'])}
-    chart_data['model_show'] = chart_data['Model'].map(lambda x: display_model_names.get(x, x))
-
-
-    models = st.multiselect("Please choose the model", 
-                            sorted(chart_data['model_show'].tolist()), 
-                            default = sorted(chart_data['model_show'].tolist()),
-                            )
-    
-    chart_data = chart_data[chart_data['model_show'].isin(models)]
-    chart_data = chart_data.sort_values(by=[displayname], ascending=cus_sort).dropna(axis=0)
-
-    if len(chart_data) == 0: return
-    
-    
-    # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
-    '''
-    Show Table
-    '''
-    with st.container():
-        st.markdown('##### TABLE')
-
-        
-        model_link = {key.strip(): val for key, val in zip(info_df['Proper Display Name'], info_df['Link'])}
-
-        chart_data['model_link'] = chart_data['model_show'].map(model_link) 
-
-        chart_data_table = chart_data[['model_show', chart_data.columns[1], chart_data.columns[3]]]
-
-        # Format numeric columns to 2 decimal places
-        #chart_data_table[chart_data_table.columns[1]] = chart_data_table[chart_data_table.columns[1]].apply(lambda x: round(float(x), 3) if isinstance(float(x), (int, float)) else float(x))
-        cur_dataset_name = chart_data_table.columns[1]
-
-
-        def highlight_first_element(x):
-                # Create a DataFrame with the same shape as the input
-                df_style = pd.DataFrame('', index=x.index, columns=x.columns)
-                # Apply background color to the first element in row 0 (df[0][0])
-                # df_style.iloc[0, 1] = 'background-color: #b0c1d7; color: white'
-                df_style.iloc[0, 1] = 'background-color: #b0c1d7'
-                
-                return df_style
-
-        if cur_dataset_name in wer_displayname2datasetname:
-            chart_data_table = chart_data_table.sort_values(
-                    by=chart_data_table.columns[1],
-                    ascending=True
-                ).reset_index(drop=True)
-        else:
-            chart_data_table = chart_data_table.sort_values(
-                    by=chart_data_table.columns[1],
-                    ascending=False
-                ).reset_index(drop=True)
-            
-
-        styled_df = chart_data_table.style.format(
-            {chart_data_table.columns[1]: "{:.3f}"}
-        ).apply(
-            highlight_first_element, axis=None
-        )
-
-
-        st.dataframe(
-                styled_df,
-                column_config={
-                    'model_show': 'Model',
-                    chart_data_table.columns[1]: {'alignment': 'left'},
-                    "model_link": st.column_config.LinkColumn(
-                        "Model Link",
-                    ),
-                },
-                hide_index=True,
-                use_container_width=True
-            )
-        
-
-    # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
-    '''
-    Show Chart
-    '''
-
-    # Initialize a session state variable for toggling the chart visibility
-    if "show_chart" not in st.session_state:
-        st.session_state.show_chart = False
-
-    # Create a button to toggle visibility
-    if st.button("Show Chart"):
-        st.session_state.show_chart = not st.session_state.show_chart
-
-    if st.session_state.show_chart:
-
-        with st.container():
-            st.markdown('##### CHART')
-
-            # Get Values
-            data_values = chart_data.iloc[:, 1]
-            
-            # Calculate Q1 and Q3
-            q1 = data_values.quantile(0.25)
-            q3 = data_values.quantile(0.75)
-
-            # Calculate IQR
-            iqr = q3 - q1
-
-            # Define lower and upper bounds (1.5*IQR is a common threshold)
-            lower_bound = q1 - 1.5 * iqr
-            upper_bound = q3 + 1.5 * iqr
-
-            # Filter data within the bounds
-            filtered_data = data_values[(data_values >= lower_bound) & (data_values <= upper_bound)]
-
-            # Calculate min and max values after outlier handling
-            min_value = round(filtered_data.min() - 0.1 * filtered_data.min(), 3)
-            max_value = round(filtered_data.max() + 0.1 * filtered_data.max(), 3)
-
-            options = {
-                # "title": {"text": f"{dataset_name}"},
-                "tooltip": {
-                    "trigger": "axis",
-                    "axisPointer": {"type": "cross", "label": {"backgroundColor": "#6a7985"}},
-                    "triggerOn": 'mousemove',
-                },
-                "legend": {"data": ['Overall Accuracy']},
-                "toolbox": {"feature": {"saveAsImage": {}}},
-                "grid": {"left": "3%", "right": "4%", "bottom": "3%", "containLabel": True},
-                "xAxis": [
-                    {
-                        "type": "category",
-                        "boundaryGap": True,
-                        "triggerEvent": True,
-                        "data":  chart_data['model_show'].tolist(),
-                    }
-                ],
-                "yAxis": [{"type": "value", 
-                            "min": min_value,
-                            "max": max_value, 
-                            "boundaryGap": True
-                            # "splitNumber": 10
-                            }],
-                "series": [{
-                        "name": f"{dataset_name}",
-                        "type": "bar",
-                        "data": chart_data[f'{displayname}'].tolist(),
-                    }],
-            }
-            
-            events = {
-                "click": "function(params) { return params.value }"
-            }
-
-            value = st_echarts(options=options, events=events, height="500px")
-            

app/pages.py

@@ -1,576 +0,0 @@
-import streamlit as st
-from app.draw_diagram import *
-from app.content import *
-from app.summarization import *
-from app.show_examples import *
-
-def dataset_contents(dataset, metrics):
-    
-    custom_css = """
-                <style>
-                .my-dataset-info {
-                # background-color: #F9EBEA;
-                # padding: 10px;
-                color: #050505;
-                font-style: normal;
-                font-size: 8px;
-                height: auto;
-                }
-                </style>
-                """
-    st.markdown(custom_css, unsafe_allow_html=True)
-    st.markdown(f"""<div class="my-dataset-info">
-                    <p><b>About this dataset</b>:</p>
-                    <p>{dataset}</p>
-                    </div>""", unsafe_allow_html=True)
-    st.markdown(f"""<div class="my-dataset-info">
-                    <p><b>About this metric</b>:</p>
-                    <p>{metrics}</p>
-                    </div>""", unsafe_allow_html=True)
-
-
-def dashboard():
-
-    with st.container():
-        st.title("Leaderboard for AudioBench")
-   
-        st.markdown("""
-            [gh1]: https://github.com/AudioLLMs/AudioBench
-            [gh2]: https://github.com/AudioLLMs/AudioBench
-            **Toolkit:** [![GitHub Repo stars](https://img.shields.io/github/stars/AudioLLMs/AudioBench?style=social)][gh1] | 
-            [**Paper @ NAACL 2025**](https://arxiv.org/abs/2406.16020) | 
-            **Resource for AudioLLMs:** [![GitHub Repo stars](https://img.shields.io/github/stars/AudioLLMs/Awesome-Audio-LLM?style=social)][gh2]
-            """)
-
-
-    st.markdown("""
-            #### Recent updates
-            - **Nov. 2025**: Added Omnilingual-LLM-ASR-7B and MERaLiON-SpeechEncoder2-ASR-CTC to the leaderboard.
-            - **May. 2025**: Further expanded ASR task to 7 different languages. Added MERaLiON-2 to the leaderboard.
-            - **Jan. 2025**: AudioBench is officially accepted to NAACL 2025!
-            - **Jan. 2025**: Update the layout.
-            - **Dec. 2024**: Added MuChoMusic dataset for Music Understanding - MCQ Questions. From Paper: https://arxiv.org/abs/2408.01337.
-            - **Dec. 2024**: Singlish ASR task added! The datasets are available on [HF](https://huggingface.co/datasets/MERaLiON/MNSC).
-            - **Dec. 2024**: Updated layout and added support for comparison between models with similar sizes. 1) Reorganized layout for a better user experience. 2) Added performance summary for each task.
-            - **Aug. 2024**: Initial leaderboard is now online.
-            """)
-
-    st.divider()
-    
-    st.markdown("""
-                #### Evaluating Audio-based Large Language Models
-                
-                - AudioBench is a comprehensive evaluation benchmark designed for general instruction-following audio large language models.
-                - AudioBench is an evaluation benchmark that we continually improve and maintain.
-                
-                Below are the initial 26 datasets that are included in AudioBench. We are now exteneded to over 40 datasets and going to extend to more in the future.
-                """
-                )
-
-
-    with st.container():
-        
-        st.markdown('''
-                ''')
-        
-        st.markdown("###### :dart: Our Benchmark includes: ")
-        cols = st.columns(8)
-        cols[0].metric(label="Tasks", value=">8")
-        cols[1].metric(label="Datasets", value=">40")
-        cols[2].metric(label="Evaluated Models", value=">5")
-    
-    st.divider()
-    with st.container():
-        left_co, right_co = st.columns([1, 0.1])
-
-        with left_co:
-            st.markdown("""
-                        ##### Citations :round_pushpin:
-                        ```
-                        @article{wang2024audiobench,
-                            title={AudioBench: A Universal Benchmark for Audio Large Language Models},
-                            author={Wang, Bin and Zou, Xunlong and Lin, Geyu and Sun, Shuo and Liu, Zhuohan and Zhang, Wenyu and Liu, Zhengyuan and Aw, AiTi and Chen, Nancy F},
-                            journal={NAACL},
-                            year={2025}
-                            }
-                        ```
-                        ```
-                        @article{zhang2024mowe,
-                            title={MoWE-Audio: Multitask AudioLLMs with Mixture of Weak Encoders},
-                            author={Zhang, Wenyu and Sun, Shuo and Wang, Bin and Zou, Xunlong and Liu, Zhuohan and He, Yingxu and Lin, Geyu and Chen, Nancy F and Aw, Ai Ti},
-                            journal={ICASSP},
-                            year={2025}
-                            }
-                        ```
-                        ```
-                        @article{wang2025advancing,
-                            title={Advancing Singlish Understanding: Bridging the Gap with Datasets and Multimodal Models},
-                            author={Wang, Bin and Zou, Xunlong and Sun, Shuo and Zhang, Wenyu and He, Yingxu and Liu, Zhuohan and Wei, Chengwei and Chen, Nancy F and Aw, AiTi},
-                            journal={arXiv preprint arXiv:2501.01034},
-                            year={2025}
-                            }
-                        ```
-                        ```
-                        @article{he2024meralion,
-                            title={MERaLiON-AudioLLM: Technical Report},
-                            author={He, Yingxu and Liu, Zhuohan and Sun, Shuo and Wang, Bin and Zhang, Wenyu and Zou, Xunlong and Chen, Nancy F and Aw, Ai Ti},
-                            journal={arXiv preprint arXiv:2412.09818},
-                            year={2024}
-                            }
-                        ```
-
-                        """)
-
-
-def asr_english():
-    st.title("Task: Automatic Speech Recognition - English")
-    
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_english_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_english', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_english', filter_1, 'wer', cus_sort=True)
-
-
-def asr_singlish():
-    st.title("Task: Automatic Speech Recognition - Singlish")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_singlish_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_singlish', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_singlish', filter_1, 'wer')
-
-
-def asr_mandarin():
-    st.title("Task: Automatic Speech Recognition - Mandarin")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_mandarin_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_mandarin', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_mandarin', filter_1, 'wer')
-
-
-def asr_malay():
-    st.title("Task: Automatic Speech Recognition - Malay")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_malay_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_malay', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_malay', filter_1, 'wer')
-
-
-def asr_tamil():
-    st.title("Task: Automatic Speech Recognition - Tamil")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_tamil_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_tamil', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_tamil', filter_1, 'wer')
-
-
-def asr_indonesian():
-    st.title("Task: Automatic Speech Recognition - Indonesian")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_indonesian_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_indonesian', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_indonesian', filter_1, 'wer')
-
-
-def asr_thai():
-    st.title("Task: Automatic Speech Recognition - Thai")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_thai_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_thai', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_thai', filter_1, 'wer')
-
-
-def asr_vietnamese():
-    st.title("Task: Automatic Speech Recognition - Vietnamese")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_vietnamese_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_vietnamese', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_vietnamese', filter_1, 'wer')
-
-
-def asr_private():
-    st.title("Task: Automatic Speech Recognition - Other Datasets")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + asr_private_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('asr_private', ['wer'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
-            draw('su', 'asr_private', filter_1, 'wer')
-
-
-def speech_translation():
-    st.title("Task: Speech Translation")
-    
-    sum = ['Overall']
-
-    filters_levelone = sum + speech_translation_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('st', ['bleu'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['bleu'])
-            draw('su', 'ST', filter_1, 'bleu')
-
-
-def speech_question_answering_english():
-    st.title("Task: Spoken Question Answering - English")
-    
-    sum = ['Overall']
-
-    filters_levelone = sum + speech_qa_english_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('sqa_english', ['llama3_70b_judge'])
-
-        #elif filter_1 in dataset_lists:
-        #    dataset_contents(sqa_datasets[filter_1], metrics['llama3_70b_judge'])
-        #    draw('su', 'SQA', filter_1, 'llama3_70b_judge')
-        
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('su', 'sqa_english', filter_1, 'llama3_70b_judge')
-
-
-def speech_question_answering_singlish():
-    st.title("Task: Spoken Question Answering - Singlish")
-    
-    sum = ['Overall']
-
-    filters_levelone = sum + speech_qa_singlish_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left: 
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('sqa_singlish', ['llama3_70b_judge'])
-        
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('su', 'sqa_singlish', filter_1, 'llama3_70b_judge')
-
-
-def spoken_dialogue_summarization_singlish():
-    st.title("Task: Spoken Dialogue Summarization - Singlish")
-    
-    sum = ['Overall']
-
-    filters_levelone = sum + sds_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left: 
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('sds_singlish', ['llama3_70b_judge'])
-        
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('su', 'sds_singlish', filter_1, 'llama3_70b_judge')
-
-
-def speech_instruction():
-    st.title("Task: Speech Instruction")
-    
-    sum = ['Overall']
-    
-    filters_levelone = sum + si_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('speech_instruction', ['llama3_70b_judge'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('su', 'speech_instruction', filter_1, 'llama3_70b_judge')
-
-
-def audio_captioning():
-    st.title("Task: Audio Captioning")
-
-    filters_levelone = ac_datasets
-
-    filters_leveltwo = ['Llama3-70b-judge', 'Meteor']
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    with middle:
-        metric = st.selectbox('Metric', filters_leveltwo)
-
-    if filter_1 or metric:
-        dataset_contents(dataset_diaplay_information[filter_1], metrics_info[metric.lower().replace('-', '_')])
-        draw('asu', 'audio_captioning', filter_1, metric.lower().replace('-', '_'))
-
-
-def audio_scene_question_answering():
-    st.title("Task: Audio Scene Question Answering")
-
-    sum = ['Overall']
-    
-    filters_levelone = sum + asqa_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('audio_scene_question_answering', ['llama3_70b_judge'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('asu', 'audio_scene_question_answering', filter_1, 'llama3_70b_judge')
-
-
-def emotion_recognition():
-    st.title("Task: Emotion Recognition")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + er_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('emotion_recognition', ['llama3_70b_judge'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('vu', 'emotion_recognition', filter_1, 'llama3_70b_judge')
-
-
-def accent_recognition():
-    st.title("Task: Accent Recognition")
-
-    sum = ['Overall']
-
-    filters_levelone = sum + ar_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-
-
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('accent_recognition', ['llama3_70b_judge'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('vu', 'accent_recognition', filter_1, 'llama3_70b_judge')
-
-
-def gender_recognition():
-    st.title("Task: Gender Recognition")
-    
-    sum = ['Overall']
-
-    filters_levelone = sum + gr_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('gender_recognition', ['llama3_70b_judge'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('vu', 'gender_recognition', filter_1, 'llama3_70b_judge')
-
-
-def music_understanding():
-    st.title("Task: Music Understanding - MCQ Questions")
-    
-    sum = ['Overall']
-
-    filters_levelone = sum + music_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    if filter_1:
-        if filter_1 in sum:
-            sum_table_mulit_metrix('music_understanding', ['llama3_70b_judge'])
-        else:
-            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
-            draw('vu', 'music_understanding', filter_1, 'llama3_70b_judge')
-
-
-def under_development():
-    st.title("Task: Under Development")
-
-    filters_levelone = non_wer_development_datasets + wer_development_datasets
-    
-    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
-    
-    with left:
-        filter_1 = st.selectbox('Dataset', filters_levelone)
-    
-    dataset_contents(dataset_diaplay_information[filter_1], 'under_development')
-    
-    # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
-
-    '''
-    Show Dataset Examples
-    '''
-
-    # Initialize a session state variable for toggling the chart visibility
-    if "show_dataset_examples" not in st.session_state:
-        st.session_state.show_dataset_examples = False
-
-    # Create a button to toggle visibility
-    if st.button("Show Dataset Examples"):
-        st.session_state.show_dataset_examples = not st.session_state.show_dataset_examples
-
-    if st.session_state.show_dataset_examples:
-        
-        # st.markdown('To be implemented')
-
-        # # if dataset_name in ['Earnings21-Test', 'Earnings22-Test', 'Tedlium3-Test', 'Tedlium3-Long-form-Test']:
-        if filter_1 in []:
-            pass
-        else:
-            try:
-                show_dataset_examples(filter_1)
-            except:
-                st.markdown('To be implemented')
-    # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =    
-    
-    if filter_1 in wer_development_datasets:
-        draw('vu', 'under_development_wer', filter_1, 'wer')
-
-    elif filter_1 in non_wer_development_datasets:
-        draw('vu', 'under_development_llama3_70b_judge', filter_1, 'llama3_70b_judge')

app/show_examples.py

@@ -1,199 +0,0 @@
-import streamlit as st
-import datasets
-import numpy as np
-import io
-import soundfile as sf
-
-import html
-
-from app.content import displayname2datasetname
-
-def show_dataset_examples(display_name):
-    st.divider()
-    dataset_name = displayname2datasetname[display_name]
-    sample_folder = f"./examples/{dataset_name}"
-    
-    # load dataset
-    dataset = datasets.load_from_disk(sample_folder)
-    
-    for index in range(len(dataset)):
-        with st.container():
-            st.markdown(f'##### Example-{index+1}')
-            col1, col2 = st.columns([0.3, 0.7], vertical_alignment="center")
-
-            with col1:
-                # Convert the NumPy array to a WAV file in memory
-                bytes_io = io.BytesIO()
-                sf.write(bytes_io, dataset[index]['context']['audio']['array'], dataset[index]['context']['audio']['sampling_rate'], format='WAV')
-                bytes_io.seek(0)
-                # Play audio in Streamlit
-                st.audio(bytes_io, format='audio/wav')
-                # st.audio(f'{sample_folder}/sample_{index}.wav', format="audio/wav")
-                       
-            if dataset_name in ['CN-College-Listen-MCQ-Test', 'DREAM-TTS-MCQ-Test']:
-                
-                choices = dataset[index]['other_attributes']['choices'] 
-                if isinstance(choices, str):
-                    choices_text = choices
-                elif isinstance(choices, list):
-                    choices_text = ' '.join(i for i in choices)
-                
-                question_text = f"""{dataset[index]['instruction']['text']} {choices_text}"""
-            else:
-                question_text = f"""{dataset[index]['instruction']['text']}"""
-            
-            question_text = html.escape(question_text)
-        
-        # with st.container():
-        with col2:
-            custom_css = """
-                        <style>
-                        .my-container-table, p.my-container-text {
-                        background-color: #fcf8dc;
-                        padding: 10px;
-                        border-radius: 5px;
-                        font-size: 13px;
-                        # height: 50px;
-                        word-wrap: break-word
-                        }
-                        </style>
-                        """
-            st.markdown(custom_css, unsafe_allow_html=True)
-
-            body_details = f"""<table style="table-layout: fixed; width:100%">
-            <thead>
-                <tr style="text-align: center;">
-                    <th style="width:50%">PROMPT</th>
-                    <th style="width:50%">ANSWER</th>
-                </tr>
-                <tr>
-                    <td><b>{html.escape(question_text.replace('(A)', '<br>(A)').replace('(B)', '<br>(B)').replace('(C)', '<br>(C)'))}
-                    </td>
-                    <td><b>{html.escape(dataset[index]['answer']['text'])}
-                    </td>
-                </tr>
-            </thead>
-            </table>"""
-            
-            st.markdown(f"""<div class="my-container-table">
-                            {body_details}
-                            </div>""", unsafe_allow_html=True)
-        
-            st.text("")
-    
-        st.divider()
-    
-
-def show_examples(category_name, dataset_name, model_lists, display_model_names):
-    st.divider()
-    sample_folder = f"./examples/{category_name}/{dataset_name}"
-    
-    dataset = datasets.load_from_disk(sample_folder)
-
-    for index in range(len(dataset)):
-        with st.container():
-            st.markdown(f'##### Example-{index+1}')
-            col1, col2 = st.columns([0.3, 0.7], vertical_alignment="center")
-
-            # with col1:
-            st.audio(f'{sample_folder}/sample_{index}.wav', format="audio/wav")
-                        
-            if dataset_name in ['CN-College-Listen-MCQ-Test', 'DREAM-TTS-MCQ-Test']:
-                
-                choices = dataset[index]['other_attributes']['choices'] 
-                if isinstance(choices, str):
-                    choices_text = choices
-                elif isinstance(choices, list):
-                    choices_text = ' '.join(i for i in choices)
-                
-                question_text = f"""{dataset[index]['instruction']['text']} {choices_text}"""
-            else:
-                question_text = f"""{dataset[index]['instruction']['text']}"""
-
-            question_text = html.escape(question_text)
-            
-            # st.divider()
-            with st.container():
-                custom_css = """
-                            <style>
-                            .my-container-table, p.my-container-text {
-                            background-color: #fcf8dc;
-                            padding: 10px;
-                            border-radius: 5px;
-                            font-size: 13px;
-                            # height: 50px;
-                            word-wrap: break-word
-                            }
-                            </style>
-                            """
-                st.markdown(custom_css, unsafe_allow_html=True)
-
-                model_lists.sort()
-
-                s = f"""<tr>
-                       <td><b>REFERENCE</td>
-                       <td><b>{html.escape(question_text.replace('(A)', '<br>(A)').replace('(B)', '<br>(B)').replace('(C)', '<br>(C)'))}
-                       </td>
-                       <td><b>{html.escape(dataset[index]['answer']['text'])}
-                       </td>
-                </tr>
-                """
-                if dataset_name in ['CN-College-Listen-MCQ-Test', 'DREAM-TTS-MCQ-Test']:
-                    for model in model_lists:
-                        try:
-
-                            model_prediction = dataset[index][model]['model_prediction']
-                            model_prediction = model_prediction.replace('<','').replace('>','').replace('\n','(newline)').replace('*','')
-
-                            s += f"""<tr>
-                                <td>{display_model_names[model]}</td>
-                                <td>
-                                    {dataset[index][model]['text'].replace('Choices:', '<br>Choices:').replace('(A)', '<br>(A)').replace('(B)', '<br>(B)').replace('(C)', '<br>(C)') 
-                                     }
-                                    </td>
-                                <td>{html.escape(model_prediction)}</td>
-                            </tr>"""
-                        except:
-                            print(f"{model} is not in {dataset_name}")
-                            continue
-                else:
-                    for model in model_lists:
-
-                        print(dataset[index][model]['model_prediction'])
-
-                        try:
-
-                            model_prediction = dataset[index][model]['model_prediction']
-                            model_prediction = model_prediction.replace('<','').replace('>','').replace('\n','(newline)').replace('*','')
-
-                            s += f"""<tr>
-                                <td>{display_model_names[model]}</td>
-                                <td>{html.escape(dataset[index][model]['text'])}</td>
-                                <td>{html.escape(model_prediction)}</td>
-                            </tr>"""
-                        except:
-                            print(f"{model} is not in {dataset_name}")
-                            continue
-
-                
-                body_details = f"""<table style="table-layout: fixed; width:100%">
-                <thead>
-                    <tr style="text-align: center;">
-                        <th style="width:20%">MODEL</th>
-                        <th style="width:30%">QUESTION</th>
-                        <th style="width:50%">MODEL PREDICTION</th>
-                    </tr>
-                {s}
-                </thead>
-                </table>"""
-                
-                st.markdown(f"""<div class="my-container-table">
-                                {body_details}
-                                </div>""", unsafe_allow_html=True)
-            
-                st.text("")
-        
-        st.divider()
-
-    
-    

app/summarization.py

@@ -1,133 +0,0 @@
-import streamlit as st
-import pandas as pd
-import numpy as np
-from streamlit_echarts import st_echarts
-from streamlit.components.v1 import html
-# from PIL import Image 
-from app.show_examples import *
-from app.content import *
-
-import pandas as pd
-from typing import List
-
-from model_information import get_dataframe
-
-info_df = get_dataframe()
-
-
-def sum_table_mulit_metrix(task_name, metrics_lists: List[str]):
-
-    # combine chart data from multiple sources
-    chart_data = pd.DataFrame()
-    for metrics in metrics_lists:
-        folder = f"./results_organized/{metrics}"
-        data_path = f'{folder}/{task_name.lower()}.csv'
-        one_chart_data = pd.read_csv(data_path).round(3)
-        if len(chart_data) == 0:
-            chart_data = one_chart_data
-        else:
-            chart_data = pd.merge(chart_data, one_chart_data, on='Model', how='outer')
-
-    selected_columns = [i for i in chart_data.columns if i != 'Model']
-    # TODO: temp code. delete this after ytb tamil vs is fully tested. 
-    _columns_to_exclude_from_average = ["ytb_asr_batch3_tamil_v2"]
-    selected_columns = [col for col in selected_columns if col not in _columns_to_exclude_from_average]
-    chart_data['Average'] = chart_data[selected_columns].mean(axis=1)
-
-    # Update dataset name in table
-    chart_data = chart_data.rename(columns=datasetname2diaplayname)
-    
-    st.markdown("""
-                <style>
-                .stMultiSelect [data-baseweb=select] span {
-                    max-width: 800px;
-                    font-size: 0.9rem;
-                    background-color: #3C6478 !important; /* Background color for selected items */
-                    color: white; /* Change text color */
-                    back
-                }
-                </style>
-                """, unsafe_allow_html=True)
-    
-    # remap model names
-    display_model_names = {key.strip() :val.strip() for key, val in zip(info_df['Original Name'], info_df['Proper Display Name'])}
-    chart_data['model_show'] = chart_data['Model'].map(lambda x: display_model_names.get(x, x))
-
-    models = st.multiselect("Please choose the model", 
-                            sorted(chart_data['model_show'].tolist()), 
-                            default = sorted(chart_data['model_show'].tolist()),
-                            )
-    # TODO: delete this after ytb tamil v2 is fully utilized. 
-    if task_name == 'asr_tamil':
-        chart_data = chart_data[chart_data['model_show'].isin(models)]
-    else:
-        chart_data = chart_data[chart_data['model_show'].isin(models)].dropna(axis=0)
-
-    if len(chart_data) == 0: return
-
-    # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
-    '''
-    Show Table
-    '''
-    with st.container():
-        st.markdown(f'##### TABLE')
-
-        model_link = {key.strip(): val for key, val in zip(info_df['Proper Display Name'], info_df['Link'])}
-
-        chart_data['model_link'] = chart_data['model_show'].map(model_link) 
-
-        tabel_columns = [i for i in chart_data.columns if i not in ['Model', 'model_show']]
-        column_to_front = 'Average'
-        new_order = [column_to_front] + [col for col in tabel_columns if col != column_to_front]
-        
-        chart_data_table = chart_data[['model_show'] + new_order]
-        
-
-        # Format numeric columns to 2 decimal places
-        target_column = chart_data_table.columns[1]
-        chart_data_table.loc[:, target_column] = chart_data_table[target_column].apply(lambda x: round(float(x), 3) if isinstance(float(x), (int, float)) else float(x))
-
-        if metrics in ['wer']:
-            ascend = True
-        else:
-            ascend= False
-
-        chart_data_table = chart_data_table.sort_values(
-                by=['Average'],
-                ascending=ascend
-            ).reset_index(drop=True)
-        
-        # Highlight the best performing model
-        def highlight_first_element(x):
-            # Create a DataFrame with the same shape as the input
-            df_style = pd.DataFrame('', index=x.index, columns=x.columns)
-            # Apply background color to the first element in row 0 (df[0][0])
-            # df_style.iloc[0, 1] = 'background-color: #b0c1d7; color: white'
-            df_style.iloc[0, 1] = 'background-color: #b0c1d7'
-
-            return df_style
-        
-
-        styled_df = chart_data_table.style.format(
-            {
-                chart_data_table.columns[i]: "{:.3f}" for i in range(1, len(chart_data_table.columns) - 1)
-             }
-        ).apply(
-            highlight_first_element, axis=None
-        )
-
-        st.dataframe(
-                styled_df,
-                column_config={
-                    'model_show': 'Model',
-                    chart_data_table.columns[1]: {'alignment': 'left'},
-                    "model_link": st.column_config.LinkColumn(
-                        "Model Link",
-                    ),
-                },
-                hide_index=True,
-                use_container_width=True
-            )
-            
-    # Only report the last metrics
-    st.markdown(f'###### Metric: {metrics_info[metrics]}')

assets/index-D38mSSRC.css

@@ -0,0 +1 @@
+*,*:before,*:after{box-sizing:border-box;margin:0;padding:0}:root{--bg:#f4f6fa;--surface:#fff;--border:#e2e6ef;--text:#1a2332;--text2:#5a6377;--text3:#8892a4;--accent:#3C6478;--accent-light:#eaf2f7;--best:#dbeafe;--radius:10px;--shadow:0 1px 3px rgba(0,0,0,.06);--sidebar-w:260px;--font:-apple-system,BlinkMacSystemFont,"Segoe UI",Roboto,"Helvetica Neue",Arial,sans-serif}html{font-size:14px}body{font-family:var(--font);color:var(--text);background:var(--bg);line-height:1.55}a{color:var(--accent);text-decoration:none}a:hover{text-decoration:underline}.layout{display:flex;min-height:100vh}.sidebar{width:var(--sidebar-w);background:linear-gradient(180deg,#1a2332,#243447);color:#c8d1dc;position:fixed;top:0;left:0;bottom:0;overflow-y:auto;display:flex;flex-direction:column;z-index:10}.sidebar h2{padding:20px 18px 8px;font-size:1.1rem;color:#fff;letter-spacing:.02em}.sidebar-links{margin-top:12px}.main{margin-left:var(--sidebar-w);flex:1;padding:28px 36px 60px;max-width:1280px;min-width:0;overflow-x:hidden}.nav-group-label{padding:10px 18px 6px;font-size:.75rem;text-transform:uppercase;letter-spacing:.07em;color:#8a9bb0;font-weight:700;cursor:pointer;display:flex;align-items:center;gap:6px;-webkit-user-select:none;user-select:none;border-left:3px solid transparent;transition:background .12s,color .12s,border-color .12s}.nav-group-label:hover{color:#b4c5d8;background:#ffffff0a}.nav-group-label .arrow{font-size:.6rem;transition:transform .15s}.nav-group-label .arrow.open{transform:rotate(90deg)}.nav-group-label.expandable{padding-bottom:4px}.nav-group-label.clickable{padding:10px 18px}.nav-group-label.clickable:hover{background:#ffffff0f;color:#c8d6e5}.nav-group-label.clickable.active{background:#ffffff1a;color:#fff;border-left-color:var(--accent)}.nav-item.nav-dashboard{padding:7px 18px;font-size:.85rem;font-weight:600}.nav-item{display:block;padding:6px 18px 6px 32px;font-size:.82rem;color:#7e93aa;cursor:pointer;transition:background .12s,color .12s;border-left:3px solid transparent}.nav-item:hover{background:#ffffff0d;color:#b4c5d8}.nav-item.active{background:#ffffff14;color:#e8edf3;border-left-color:var(--accent);font-weight:600}.hero{background:linear-gradient(135deg,#1a2332,#2d4a5e 50%,#3c6478);border-radius:var(--radius);padding:36px 40px;color:#fff;margin-bottom:24px}.hero h1{font-size:2rem;font-weight:800;margin-bottom:6px}.hero .links{display:flex;gap:12px;flex-wrap:wrap}.hero .links a{color:#a0cfec;font-size:.88rem}.stat-row{display:flex;gap:14px;flex-wrap:wrap;margin-bottom:20px}.stat-card{background:var(--surface);border:1px solid var(--border);border-radius:var(--radius);padding:14px 20px;min-width:120px;flex:1;box-shadow:var(--shadow)}.stat-card .label{font-size:.72rem;text-transform:uppercase;letter-spacing:.04em;color:var(--text2);font-weight:700;margin-bottom:2px}.stat-card .value{font-size:1.5rem;font-weight:700;color:var(--text)}.card{background:var(--surface);border:1px solid var(--border);border-radius:var(--radius);padding:20px 24px;margin-bottom:16px;box-shadow:var(--shadow);overflow:hidden}.section-title{font-size:1rem;font-weight:700;margin-bottom:12px;color:var(--text)}.collapsible-header{cursor:pointer;display:flex;align-items:center;gap:8px;font-weight:700;font-size:.92rem;color:var(--text);-webkit-user-select:none;user-select:none;padding:12px 0}.collapsible-header:hover{color:var(--accent)}.collapsible-body{padding-bottom:14px}.table-wrap{overflow-x:auto;margin-bottom:16px}table.lb{width:100%;border-collapse:separate;border-spacing:0;font-size:.85rem}table.lb th{position:sticky;top:0;background:var(--bg);padding:10px 12px;text-align:left;font-weight:700;color:var(--text2);white-space:nowrap;border-bottom:2px solid var(--border);font-size:.78rem;text-transform:uppercase;letter-spacing:.03em}table.lb td{padding:9px 12px;border-bottom:1px solid var(--border);white-space:nowrap}table.lb tr:hover td{background:var(--accent-light)}table.lb tr.best td{background:var(--best)}table.lb td.num{text-align:right;font-variant-numeric:tabular-nums}table.lb .model-name{font-weight:600;max-width:320px;overflow:hidden;text-overflow:ellipsis}table.lb .rank{color:var(--text3);font-weight:600;width:36px;text-align:center}table.lb th.sortable{cursor:pointer;-webkit-user-select:none;user-select:none}table.lb th.sortable:hover{color:var(--accent)}table.lb th.has-stats{position:relative}.col-header-content{display:inline-flex;align-items:center;gap:2px}.sort-indicator{font-size:.7rem;color:var(--text3);opacity:.4;transition:opacity .15s}.sort-indicator.active{opacity:1;color:var(--accent)}.col-tooltip{display:none;position:absolute;left:0;top:100%;z-index:20;background:var(--surface);border:1px solid var(--border);border-radius:8px;padding:14px 16px;min-width:260px;box-shadow:0 4px 16px #0000001f;font-weight:400;text-transform:none;letter-spacing:0;font-size:.82rem;color:var(--text);line-height:1.6;white-space:normal}table.lb th.has-stats:hover .col-tooltip{display:block}.col-tooltip .tt-label{font-size:.68rem;text-transform:uppercase;letter-spacing:.05em;color:var(--text3);font-weight:700;margin-bottom:2px}.col-tooltip .tt-val{font-size:1rem;font-weight:700;color:var(--text);margin-bottom:8px}.col-tooltip .tt-desc{font-size:.8rem;color:var(--text2);margin-bottom:8px}.col-tooltip .tt-stats{display:grid;grid-template-columns:1fr 1fr;gap:4px 14px;font-size:.78rem}.col-tooltip .tt-stat-label{color:var(--text3)}.col-tooltip .tt-stat-val{font-weight:600}.col-tooltip .tt-link{margin-top:8px;font-size:.78rem}.dataset-detail-item{border-bottom:1px solid var(--border);padding:12px 0}.dataset-detail-item:last-child{border-bottom:none}.dataset-detail-header{display:flex;align-items:baseline;gap:4px;flex-wrap:wrap}.dataset-stats-row{display:flex;align-items:flex-start;gap:20px;margin-top:6px;flex-wrap:wrap}.dataset-stats-numbers{display:flex;gap:16px;flex-wrap:wrap;font-size:.78rem;color:var(--text3)}.mini-hist{flex-shrink:0}.mini-hist-bars{display:flex;align-items:flex-end;gap:3px;height:48px}.mini-hist-col{display:flex;flex-direction:column;align-items:center;width:40px}.mini-hist-bar{width:100%;background:var(--accent);border-radius:2px 2px 0 0;min-height:1px;transition:height .2s}.mini-hist-label{font-size:.58rem;color:var(--text3);margin-top:2px;white-space:nowrap}.examples-section{margin-top:8px;background:var(--bg);border-radius:8px;padding:10px 14px}.examples-header{display:flex;align-items:center;gap:10px;margin-bottom:8px}.examples-title{font-size:.78rem;font-weight:700;color:var(--text2);text-transform:uppercase;letter-spacing:.04em}.examples-nav{display:flex;gap:4px}.examples-nav-btn{width:22px;height:22px;border-radius:50%;border:1px solid var(--border);background:var(--surface);font-size:.72rem;font-weight:700;color:var(--text2);cursor:pointer;display:flex;align-items:center;justify-content:center;transition:all .12s}.examples-nav-btn:hover{border-color:var(--accent);color:var(--accent)}.examples-nav-btn.active{background:var(--accent);color:#fff;border-color:var(--accent)}.examples-body{display:flex;flex-direction:column;gap:6px}.example-row{display:flex;align-items:flex-start;gap:10px;font-size:.82rem}.example-label{font-weight:700;color:var(--text3);font-size:.72rem;text-transform:uppercase;letter-spacing:.04em;min-width:64px;padding-top:2px;flex-shrink:0}.example-text{color:var(--text2);line-height:1.5;word-break:break-word;overflow-wrap:anywhere;min-width:0;flex:1}.examples-body audio{height:32px;max-width:300px}.metric-badge{display:inline-block;background:var(--accent-light);color:var(--accent);font-size:.74rem;font-weight:700;padding:3px 10px;border-radius:20px;margin-left:8px}.mobile-menu-btn{display:none;position:fixed;top:12px;left:12px;z-index:30;width:40px;height:40px;border-radius:8px;border:none;background:var(--accent);color:#fff;font-size:1.3rem;cursor:pointer;align-items:center;justify-content:center;box-shadow:0 2px 8px #0000002e}.sidebar-overlay{display:none;position:fixed;inset:0;background:#00000073;z-index:9}@media(max-width:860px){.mobile-menu-btn{display:flex}.sidebar{width:var(--sidebar-w);transform:translate(-100%);transition:transform .25s ease}.sidebar.open{transform:translate(0)}.sidebar-overlay.open{display:block}.main{margin-left:0;padding:52px 16px 32px}}

examples/{ytb_asr_batch3_tamil/data-00000-of-00001.arrow → cna_test/example_1.wav}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:610511896b4e2b7c2ade028585ae16c3d8e4f9cdbf7767e2fd1d23005f307a89
-size 890760
+oid sha256:d671a9ef97a17dcbc93dde2bb91ac873d8f2412b6b0b76cc63b2f241a264f1ab
+size 117868

examples/{ytb_sds_batch3_chinese/data-00000-of-00001.arrow → commonvoice_17_id_asr/example_0.wav}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6118f3210caa8872b421495bb034b388a63cf8b14b6dc1e2e1f72347fea4e1f6
-size 260440
+oid sha256:2873873174ef2b47c9768834427313dc06c4787d5e2ece5559592074d1b2cac5
+size 262700

examples/{ytb_asr_batch3_chinese/data-00000-of-00001.arrow → commonvoice_17_id_asr/example_1.wav}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fca176caa14fa2d13d2fdf96c99a0e8dfec17e2b83af14b78a1d18a0d311ed1c
-size 152264
+oid sha256:49ee064f2f06e63ca0c0f328461051f8da550af602b4ba684bdc2d8fd562c1f7
+size 124460

examples/{ytb_asr_batch3_malay/data-00000-of-00001.arrow → commonvoice_17_ta_asr/example_0.wav}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9659b947c6b10c86c127690ac15403d40f9e71e541a4bb4391545d6329c7313f
-size 702736
+oid sha256:fb926eb3c960beee6ff08bca9498950bf1c34376419962c5a9029003942db15d
+size 146732

examples/commonvoice_17_ta_asr/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:daa9812c5ae8ff9c4645af3393c241a7b3b8609d210e80141da797591c0ec35e
+size 163628

examples/commonvoice_17_th_asr/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b84d3b0d0330d233bab30e235d4864fcd91a202cb1b124d1bf04340a75d3c40b
+size 109484

examples/commonvoice_17_vi_asr/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3232ae3fd1e9d8eb79cbf9ede448070b2813ea53d3a425101f9c653219f29d86
+size 104876

examples/commonvoice_17_vi_asr/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51c5e36f8ce7c09aeedf461accbd39fdda7bef172d24abf4c54a5d59031692ac
+size 145196

examples/commonvoice_zh_asr/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99da77bb3103c130a65cfbf6806bd368ffa8db63793c543c1743932335220b00
+size 231980

examples/commonvoice_zh_asr/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8ca40944a0d287bdc966f9f54df3104f493ade7b8bf7105217429a86c59c1b7
+size 176300

examples/fleurs_tamil_ta_30_asr/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6457a73e386ab3654f79548791b6ff73292e5a315654e266ea886ab068cc0781
+size 480044

examples/fleurs_tamil_ta_30_asr/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9561b9a63d66cc4fe3271a0ea7028ff005027080081ce1eabc554209ff180e6
+size 441644

examples/gigaspeech2_id_test/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bfec7fba413fc8e7da61d5c24d73bdd77e016b297436ac51bb8a624c949bd8ac
+size 220266

examples/gigaspeech2_th_test/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0982da6bb864511312c1589a125866e6fb44f06197c8b42771caceb99e6118ae
+size 295722

examples/gigaspeech2_th_test/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b347da5f01c383d5ab33d530a08824bb1dfa3a5ca9fddc31e0091825a6f079ce
+size 150540

examples/gigaspeech2_vi_test/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58f881efda72405f79eb391f17866d34004a124e96fc85653c928fd8d8a5370c
+size 355812

examples/gigaspeech2_vi_test/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c0a164361ecb493b76cbb26d5747b4dfc2a4d6623067b677d6ad108991bc6f0
+size 250382

examples/idpc_short_test/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c8d813d57338bb5db29d1b6ec2f428c94265da032f4feba06df05690c0b5ac2
+size 846124

examples/idpc_short_test/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad8dad30f3dddeaad7e32336841a2cb2a15d3761c38dd62ad2ec6f49d956dd68
+size 418284

examples/imda_ar_dialogue/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a25e546945a11abecef9ea36f6b4ab34d4949803cad06faf3f45857536dceba2
+size 897644

examples/imda_ar_dialogue/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91a6b03725839579b6806bd9a28057a51e8a9cd28f359fe09cfa321e6f41d4ef
+size 867124

examples/imda_ar_sentence/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e60ddc2b12238cbcad5dcc611754ab995ab8cd5789ef3ab13a095c9340e5c17
+size 158124

examples/imda_ar_sentence/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3235a16d5654c470db58d787b54d4b7b79275d4c6566007ec80ee6693ba1a2a
+size 162444

examples/imda_part1_asr_test/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:137d9c12879d7432c8665ad938e456027d621a40c3b13d4dd6f715a50177602d
+size 127724

examples/imda_part1_asr_test/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:150e574c811a6bed67870e260250383a9821db884c2931a1cdee2c4c8e165f0f
+size 176812

examples/imda_part2_asr_test/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb4bf2857925ed079a40a542bc9382154a02ccf08c4e4e0a037eeaf67ed62c3f
+size 131468

examples/imda_part2_asr_test/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a23f686148140282b7b39002f81ec9df7142d9492bb0c8cc7da84c0b51cccfbb
+size 179596

examples/imda_part3_30s_asr_test/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:125993396252ba54433006e3cf3e248b4c15e87e4a2d041f5853d9ddbd8d7cc7
+size 929644

examples/imda_part3_30s_asr_test/example_1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9834aca6dd6a7783589f7f437aeff2848d0cd4125b236fd3735eb674f880a610
+size 826828

examples/imda_part3_30s_ds_human_test/example_0.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57c2c8c700f4469ad1673408ff10c7a23eac1e206bfc6daf45597ef5176daeea
+size 955596