Spaces:

MedAiHealth
/

asr-test

Sleeping

App Files Files Community

Himel commited on Jul 12, 2024

Commit

0599b65

1 Parent(s): 5f661ec

update

Browse files

This view is limited to 50 files because it contains too many changes. See raw diff

Files changed (50) hide show

.idea/.gitignore +3 -0
.idea/asr-test.iml +10 -0
.idea/inspectionProfiles/Project_Default.xml +136 -0
.idea/inspectionProfiles/profiles_settings.xml +6 -0
.idea/misc.xml +7 -0
.idea/modules.xml +8 -0
.idea/vcs.xml +6 -0
app.py +38 -13
test_sample/2022_05_23-05_20_09.544938_file_file.wav +0 -0
test_sample/2022_06_28-12_01_43.552306_file_file.wav +0 -0
test_sample/2022_06_28-17_12_43.524460_file_file.wav +0 -0
test_sample/2022_06_28-17_49_39.887762_file_file.wav +0 -0
test_sample/2022_06_28-18_21_58.444725_file_file.wav +0 -0
test_sample/2022_08_12-09_24_04.719303_file_file.wav +0 -0
test_sample/2022_08_20-09_51_45.786068_file_file.wav +0 -0
test_sample/2023_01_07-18_27_19.712794_file_file.wav +0 -0
test_sample/2023_01_08-07_31_45.978198_file_file.wav +0 -0
test_sample/2023_01_08-08_10_06.068597_file_file.wav +0 -0
test_sample/2023_01_08-08_29_40.932774_file_file.wav +0 -0
test_sample/2023_01_08-12_28_25.997920_file_file.wav +0 -0
test_sample/2023_01_08-20_29_39.120658_file_file.wav +0 -0
test_sample/2023_01_09-03_48_06.242548_file_file.wav +0 -0
test_sample/2023_01_09-16_31_24.687076_file_file.wav +0 -0
test_sample/2023_01_09-17_44_35.401574_file_file.wav +0 -0
test_sample/2023_01_09-18_00_55.258972_file_file.wav +0 -0
test_sample/2023_01_11-18_13_20.938188_file_file.wav +0 -0
test_sample/2023_01_12-10_26_35.292442_file_file.wav +0 -0
test_sample/2023_01_21-13_00_39.729777_file_file.wav +0 -0
test_sample/2023_01_21-14_33_17.981865_file_file.wav +0 -0
test_sample/2023_01_22-06_01_11.350828_file_file.wav +0 -0
test_sample/2023_01_22-12_44_52.548517_file_file.wav +0 -0
test_sample/2023_01_22-16_11_54.653583_file_file.wav +0 -0
test_sample/2023_01_23-09_53_23.863114_file_file.wav +0 -0
test_sample/2023_01_23-10_29_20.715050_file_file.wav +0 -0
test_sample/2023_01_23-10_46_13.330743_file_file.wav +0 -0
test_sample/2023_01_25-08_23_48.788537_file_file.wav +0 -0
test_sample/2023_01_25-18_09_11.098808_file_file.wav +0 -0
test_sample/2023_01_26-06_09_56.551890_file_file.wav +0 -0
test_sample/2023_01_31-04_57_39.844268_file_file.wav +0 -0
test_sample/2023_02_05-20_33_11.645244_file_file.wav +0 -0
test_sample/2023_02_16-18_58_28.709781_file_file.wav +0 -0
test_sample/2023_02_17-05_04_12.901097_file_file.wav +0 -0
test_sample/2023_02_19-18_05_10.510147_file_file.wav +0 -0
test_sample/2023_02_19-18_11_18.278859_file_file.wav +0 -0
test_sample/2023_02_19-19_52_46.335251_file_file.wav +0 -0
test_sample/2023_02_19-21_05_48.788175_file_file.wav +0 -0
test_sample/2023_02_20-05_58_04.804689_file_file.wav +0 -0
test_sample/2023_02_21-16_15_49.133971_file_file.wav +0 -0
test_sample/2023_02_21-16_17_04.681508_file_file.wav +0 -0
test_sample/2023_02_21-18_15_05.508112_file_file.wav +0 -0

.idea/.gitignore ADDED Viewed

	@@ -0,0 +1,3 @@

+# Default ignored files
+/shelf/
+/workspace.xml

.idea/asr-test.iml ADDED Viewed

	@@ -0,0 +1,10 @@

+<?xml version="1.0" encoding="UTF-8"?>
+<module type="PYTHON_MODULE" version="4">
+  <component name="NewModuleRootManager">
+    <content url="file://$MODULE_DIR$">
+      <excludeFolder url="file://$MODULE_DIR$/venv" />
+    </content>
+    <orderEntry type="jdk" jdkName="Python 3.8 (asr-test)" jdkType="Python SDK" />
+    <orderEntry type="sourceFolder" forTests="false" />
+  </component>
+</module>

.idea/inspectionProfiles/Project_Default.xml ADDED Viewed

	@@ -0,0 +1,136 @@

+<component name="InspectionProjectProfileManager">
+  <profile version="1.0">
+    <option name="myName" value="Project Default" />
+    <inspection_tool class="PyPackageRequirementsInspection" enabled="true" level="WARNING" enabled_by_default="true">
+      <option name="ignoredPackages">
+        <value>
+          <list size="123">
+            <item index="0" class="java.lang.String" itemvalue="pydantic_core" />
+            <item index="1" class="java.lang.String" itemvalue="email_validator" />
+            <item index="2" class="java.lang.String" itemvalue="typing_extensions" />
+            <item index="3" class="java.lang.String" itemvalue="altgraph" />
+            <item index="4" class="java.lang.String" itemvalue="httpx" />
+            <item index="5" class="java.lang.String" itemvalue="scikit-learn" />
+            <item index="6" class="java.lang.String" itemvalue="nvidia-cuda-cupti-cu12" />
+            <item index="7" class="java.lang.String" itemvalue="PyYAML" />
+            <item index="8" class="java.lang.String" itemvalue="nvidia-cufft-cu12" />
+            <item index="9" class="java.lang.String" itemvalue="pickleshare" />
+            <item index="10" class="java.lang.String" itemvalue="executing" />
+            <item index="11" class="java.lang.String" itemvalue="markdown-it-py" />
+            <item index="12" class="java.lang.String" itemvalue="ftfy" />
+            <item index="13" class="java.lang.String" itemvalue="Pygments" />
+            <item index="14" class="java.lang.String" itemvalue="starlette" />
+            <item index="15" class="java.lang.String" itemvalue="nvidia-cuda-runtime-cu12" />
+            <item index="16" class="java.lang.String" itemvalue="uvicorn" />
+            <item index="17" class="java.lang.String" itemvalue="comm" />
+            <item index="18" class="java.lang.String" itemvalue="pydantic" />
+            <item index="19" class="java.lang.String" itemvalue="transformers" />
+            <item index="20" class="java.lang.String" itemvalue="pexpect" />
+            <item index="21" class="java.lang.String" itemvalue="click" />
+            <item index="22" class="java.lang.String" itemvalue="ipykernel" />
+            <item index="23" class="java.lang.String" itemvalue="contourpy" />
+            <item index="24" class="java.lang.String" itemvalue="psutil" />
+            <item index="25" class="java.lang.String" itemvalue="pyinstaller-hooks-contrib" />
+            <item index="26" class="java.lang.String" itemvalue="jedi" />
+            <item index="27" class="java.lang.String" itemvalue="nvidia-cublas-cu12" />
+            <item index="28" class="java.lang.String" itemvalue="pure-eval" />
+            <item index="29" class="java.lang.String" itemvalue="regex" />
+            <item index="30" class="java.lang.String" itemvalue="asttokens" />
+            <item index="31" class="java.lang.String" itemvalue="platformdirs" />
+            <item index="32" class="java.lang.String" itemvalue="matplotlib" />
+            <item index="33" class="java.lang.String" itemvalue="httpcore" />
+            <item index="34" class="java.lang.String" itemvalue="idna" />
+            <item index="35" class="java.lang.String" itemvalue="networkx" />
+            <item index="36" class="java.lang.String" itemvalue="nvidia-nvjitlink-cu12" />
+            <item index="37" class="java.lang.String" itemvalue="nvidia-cusparse-cu12" />
+            <item index="38" class="java.lang.String" itemvalue="numpy" />
+            <item index="39" class="java.lang.String" itemvalue="requests" />
+            <item index="40" class="java.lang.String" itemvalue="nvidia-nccl-cu12" />
+            <item index="41" class="java.lang.String" itemvalue="sniffio" />
+            <item index="42" class="java.lang.String" itemvalue="seaborn" />
+            <item index="43" class="java.lang.String" itemvalue="stack-data" />
+            <item index="44" class="java.lang.String" itemvalue="zipp" />
+            <item index="45" class="java.lang.String" itemvalue="mdurl" />
+            <item index="46" class="java.lang.String" itemvalue="nest-asyncio" />
+            <item index="47" class="java.lang.String" itemvalue="tenacity" />
+            <item index="48" class="java.lang.String" itemvalue="prompt-toolkit" />
+            <item index="49" class="java.lang.String" itemvalue="xgboost" />
+            <item index="50" class="java.lang.String" itemvalue="websockets" />
+            <item index="51" class="java.lang.String" itemvalue="annotated-types" />
+            <item index="52" class="java.lang.String" itemvalue="importlib_metadata" />
+            <item index="53" class="java.lang.String" itemvalue="scipy" />
+            <item index="54" class="java.lang.String" itemvalue="watchfiles" />
+            <item index="55" class="java.lang.String" itemvalue="tornado" />
+            <item index="56" class="java.lang.String" itemvalue="nvidia-cudnn-cu12" />
+            <item index="57" class="java.lang.String" itemvalue="plotly" />
+            <item index="58" class="java.lang.String" itemvalue="torch" />
+            <item index="59" class="java.lang.String" itemvalue="python-multipart" />
+            <item index="60" class="java.lang.String" itemvalue="pandas" />
+            <item index="61" class="java.lang.String" itemvalue="mpmath" />
+            <item index="62" class="java.lang.String" itemvalue="debugpy" />
+            <item index="63" class="java.lang.String" itemvalue="pytz" />
+            <item index="64" class="java.lang.String" itemvalue="ujson" />
+            <item index="65" class="java.lang.String" itemvalue="traitlets" />
+            <item index="66" class="java.lang.String" itemvalue="protobuf" />
+            <item index="67" class="java.lang.String" itemvalue="emoji" />
+            <item index="68" class="java.lang.String" itemvalue="joblib" />
+            <item index="69" class="java.lang.String" itemvalue="threadpoolctl" />
+            <item index="70" class="java.lang.String" itemvalue="huggingface-hub" />
+            <item index="71" class="java.lang.String" itemvalue="python-dateutil" />
+            <item index="72" class="java.lang.String" itemvalue="python-dotenv" />
+            <item index="73" class="java.lang.String" itemvalue="h11" />
+            <item index="74" class="java.lang.String" itemvalue="cycler" />
+            <item index="75" class="java.lang.String" itemvalue="MarkupSafe" />
+            <item index="76" class="java.lang.String" itemvalue="httptools" />
+            <item index="77" class="java.lang.String" itemvalue="fsspec" />
+            <item index="78" class="java.lang.String" itemvalue="nvidia-cusolver-cu12" />
+            <item index="79" class="java.lang.String" itemvalue="nvidia-curand-cu12" />
+            <item index="80" class="java.lang.String" itemvalue="filelock" />
+            <item index="81" class="java.lang.String" itemvalue="pyzmq" />
+            <item index="82" class="java.lang.String" itemvalue="safetensors" />
+            <item index="83" class="java.lang.String" itemvalue="sentencepiece" />
+            <item index="84" class="java.lang.String" itemvalue="certifi" />
+            <item index="85" class="java.lang.String" itemvalue="anyio" />
+            <item index="86" class="java.lang.String" itemvalue="pyparsing" />
+            <item index="87" class="java.lang.String" itemvalue="sympy" />
+            <item index="88" class="java.lang.String" itemvalue="uvloop" />
+            <item index="89" class="java.lang.String" itemvalue="tokenizers" />
+            <item index="90" class="java.lang.String" itemvalue="dnspython" />
+            <item index="91" class="java.lang.String" itemvalue="nvidia-cuda-nvrtc-cu12" />
+            <item index="92" class="java.lang.String" itemvalue="triton" />
+            <item index="93" class="java.lang.String" itemvalue="jupyter_client" />
+            <item index="94" class="java.lang.String" itemvalue="kiwisolver" />
+            <item index="95" class="java.lang.String" itemvalue="orjson" />
+            <item index="96" class="java.lang.String" itemvalue="fonttools" />
+            <item index="97" class="java.lang.String" itemvalue="backcall" />
+            <item index="98" class="java.lang.String" itemvalue="duckdb" />
+            <item index="99" class="java.lang.String" itemvalue="nvidia-nvtx-cu12" />
+            <item index="100" class="java.lang.String" itemvalue="charset-normalizer" />
+            <item index="101" class="java.lang.String" itemvalue="shellingham" />
+            <item index="102" class="java.lang.String" itemvalue="matplotlib-inline" />
+            <item index="103" class="java.lang.String" itemvalue="ptyprocess" />
+            <item index="104" class="java.lang.String" itemvalue="wcwidth" />
+            <item index="105" class="java.lang.String" itemvalue="jupyter_core" />
+            <item index="106" class="java.lang.String" itemvalue="Jinja2" />
+            <item index="107" class="java.lang.String" itemvalue="urllib3" />
+            <item index="108" class="java.lang.String" itemvalue="pyinstaller" />
+            <item index="109" class="java.lang.String" itemvalue="fastapi-cli" />
+            <item index="110" class="java.lang.String" itemvalue="six" />
+            <item index="111" class="java.lang.String" itemvalue="typer" />
+            <item index="112" class="java.lang.String" itemvalue="importlib_resources" />
+            <item index="113" class="java.lang.String" itemvalue="faiss-cpu" />
+            <item index="114" class="java.lang.String" itemvalue="parso" />
+            <item index="115" class="java.lang.String" itemvalue="tzdata" />
+            <item index="116" class="java.lang.String" itemvalue="ipython" />
+            <item index="117" class="java.lang.String" itemvalue="rich" />
+            <item index="118" class="java.lang.String" itemvalue="sentence-transformers" />
+            <item index="119" class="java.lang.String" itemvalue="packaging" />
+            <item index="120" class="java.lang.String" itemvalue="tqdm" />
+            <item index="121" class="java.lang.String" itemvalue="fastapi" />
+            <item index="122" class="java.lang.String" itemvalue="pillow" />
+          </list>
+        </value>
+      </option>
+    </inspection_tool>
+  </profile>
+</component>

.idea/inspectionProfiles/profiles_settings.xml ADDED Viewed

	@@ -0,0 +1,6 @@

+<component name="InspectionProjectProfileManager">
+  <settings>
+    <option name="USE_PROJECT_PROFILE" value="false" />
+    <version value="1.0" />
+  </settings>
+</component>

.idea/misc.xml ADDED Viewed

	@@ -0,0 +1,7 @@

+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="Black">
+    <option name="sdkName" value="Python 3.8 (asr-test)" />
+  </component>
+  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.8 (asr-test)" project-jdk-type="Python SDK" />
+</project>

.idea/modules.xml ADDED Viewed

	@@ -0,0 +1,8 @@

+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="ProjectModuleManager">
+    <modules>
+      <module fileurl="file://$PROJECT_DIR$/.idea/asr-test.iml" filepath="$PROJECT_DIR$/.idea/asr-test.iml" />
+    </modules>
+  </component>
+</project>

.idea/vcs.xml ADDED Viewed

	@@ -0,0 +1,6 @@

+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="VcsDirectoryMappings">
+    <mapping directory="" vcs="Git" />
+  </component>
+</project>

app.py CHANGED Viewed

@@ -1,25 +1,50 @@
 import gradio as gr
-# import os
-# os.environ.get("HF_REPO_ID")
-# Define the function that will process the audio input
-def process_audio(audio):
-    # Here you can add your audio processing code
-    return f"Received audio file: {audio}"
 # Create a list of example audio files
-examples = [
-    ["https://huggingface.co/MedAiHealth/BanglaASR-MedAI/resolve/main/wav/sample_1.wav"],
-    ["https://huggingface.co/MedAiHealth/BanglaASR-MedAI/resolve/main/wav/sample_2.wav"],
-    ["https://huggingface.co/MedAiHealth/BanglaASR-MedAI/resolve/main/wav/sample_3.wav"]
-]
 # Create the Gradio interface
 interface = gr.Interface(
-    fn=process_audio,
     inputs=gr.Audio(sources="microphone", type="filepath"),
     outputs=gr.Textbox(),
     examples=examples
 )
 # Launch the interface
-interface.launch()

+import os
 import gradio as gr
+import librosa
+import torch
+import torchaudio
+import numpy as np
+from transformers import WhisperTokenizer
+from transformers import WhisperProcessor
+from transformers import WhisperFeatureExtractor
+from transformers import WhisperForConditionalGeneration
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+model_path = os.environ.get("HF_REPO_ID")
+access_token = os.environ.get("HF_TOKEN")
+feature_extractor = WhisperFeatureExtractor.from_pretrained(model_path, token=access_token)
+tokenizer = WhisperTokenizer.from_pretrained(model_path, token=access_token)
+processor = WhisperProcessor.from_pretrained(model_path, token=access_token)
+model = WhisperForConditionalGeneration.from_pretrained(model_path, token=access_token).to(device)
+def transcribe_audio(file_path):
+    speech_array, sampling_rate = torchaudio.load(file_path, format="wav")
+    speech_array = speech_array[0].numpy()
+    speech_array = librosa.resample(np.asarray(speech_array), orig_sr=sampling_rate, target_sr=16000)
+    input_features = feature_extractor(speech_array, sampling_rate=16000, return_tensors="pt").input_features
+    # batch = processor.feature_extractor.pad(input_features, return_tensors="pt")
+    predicted_ids = model.generate(inputs=input_features.to(device))[0]
+    transcription = processor.decode(predicted_ids, skip_special_tokens=True)
+    return transcription
 # Create a list of example audio files
+examples = [f"test_sample/{x}" for x in os.listdir("test_sample")]
 # Create the Gradio interface
 interface = gr.Interface(
+    fn=transcribe_audio,
     inputs=gr.Audio(sources="microphone", type="filepath"),
     outputs=gr.Textbox(),
     examples=examples
 )
 # Launch the interface
+interface.launch()

test_sample/2022_05_23-05_20_09.544938_file_file.wav ADDED Viewed

Binary file (307 kB). View file

test_sample/2022_06_28-12_01_43.552306_file_file.wav ADDED Viewed

Binary file (317 kB). View file

test_sample/2022_06_28-17_12_43.524460_file_file.wav ADDED Viewed

Binary file (219 kB). View file

test_sample/2022_06_28-17_49_39.887762_file_file.wav ADDED Viewed

Binary file (286 kB). View file

test_sample/2022_06_28-18_21_58.444725_file_file.wav ADDED Viewed

Binary file (230 kB). View file

test_sample/2022_08_12-09_24_04.719303_file_file.wav ADDED Viewed

Binary file (242 kB). View file

test_sample/2022_08_20-09_51_45.786068_file_file.wav ADDED Viewed

Binary file (219 kB). View file

test_sample/2023_01_07-18_27_19.712794_file_file.wav ADDED Viewed

Binary file (165 kB). View file

test_sample/2023_01_08-07_31_45.978198_file_file.wav ADDED Viewed

Binary file (215 kB). View file

test_sample/2023_01_08-08_10_06.068597_file_file.wav ADDED Viewed

Binary file (206 kB). View file

test_sample/2023_01_08-08_29_40.932774_file_file.wav ADDED Viewed

Binary file (230 kB). View file

test_sample/2023_01_08-12_28_25.997920_file_file.wav ADDED Viewed

Binary file (229 kB). View file

test_sample/2023_01_08-20_29_39.120658_file_file.wav ADDED Viewed

Binary file (263 kB). View file

test_sample/2023_01_09-03_48_06.242548_file_file.wav ADDED Viewed

Binary file (280 kB). View file

test_sample/2023_01_09-16_31_24.687076_file_file.wav ADDED Viewed

Binary file (209 kB). View file

test_sample/2023_01_09-17_44_35.401574_file_file.wav ADDED Viewed

Binary file (156 kB). View file

test_sample/2023_01_09-18_00_55.258972_file_file.wav ADDED Viewed

Binary file (259 kB). View file

test_sample/2023_01_11-18_13_20.938188_file_file.wav ADDED Viewed

Binary file (328 kB). View file

test_sample/2023_01_12-10_26_35.292442_file_file.wav ADDED Viewed

Binary file (215 kB). View file

test_sample/2023_01_21-13_00_39.729777_file_file.wav ADDED Viewed

Binary file (181 kB). View file

test_sample/2023_01_21-14_33_17.981865_file_file.wav ADDED Viewed

Binary file (194 kB). View file

test_sample/2023_01_22-06_01_11.350828_file_file.wav ADDED Viewed

Binary file (186 kB). View file

test_sample/2023_01_22-12_44_52.548517_file_file.wav ADDED Viewed

Binary file (181 kB). View file

test_sample/2023_01_22-16_11_54.653583_file_file.wav ADDED Viewed

Binary file (265 kB). View file

test_sample/2023_01_23-09_53_23.863114_file_file.wav ADDED Viewed

Binary file (254 kB). View file

test_sample/2023_01_23-10_29_20.715050_file_file.wav ADDED Viewed

Binary file (323 kB). View file

test_sample/2023_01_23-10_46_13.330743_file_file.wav ADDED Viewed

Binary file (196 kB). View file

test_sample/2023_01_25-08_23_48.788537_file_file.wav ADDED Viewed

Binary file (146 kB). View file

test_sample/2023_01_25-18_09_11.098808_file_file.wav ADDED Viewed

Binary file (217 kB). View file

test_sample/2023_01_26-06_09_56.551890_file_file.wav ADDED Viewed

Binary file (202 kB). View file

test_sample/2023_01_31-04_57_39.844268_file_file.wav ADDED Viewed

Binary file (184 kB). View file

test_sample/2023_02_05-20_33_11.645244_file_file.wav ADDED Viewed

Binary file (159 kB). View file

test_sample/2023_02_16-18_58_28.709781_file_file.wav ADDED Viewed

Binary file (182 kB). View file

test_sample/2023_02_17-05_04_12.901097_file_file.wav ADDED Viewed

Binary file (196 kB). View file

test_sample/2023_02_19-18_05_10.510147_file_file.wav ADDED Viewed

Binary file (277 kB). View file

test_sample/2023_02_19-18_11_18.278859_file_file.wav ADDED Viewed

Binary file (267 kB). View file

test_sample/2023_02_19-19_52_46.335251_file_file.wav ADDED Viewed

Binary file (171 kB). View file

test_sample/2023_02_19-21_05_48.788175_file_file.wav ADDED Viewed

Binary file (156 kB). View file

test_sample/2023_02_20-05_58_04.804689_file_file.wav ADDED Viewed

Binary file (227 kB). View file

test_sample/2023_02_21-16_15_49.133971_file_file.wav ADDED Viewed

Binary file (194 kB). View file

test_sample/2023_02_21-16_17_04.681508_file_file.wav ADDED Viewed

Binary file (323 kB). View file

test_sample/2023_02_21-18_15_05.508112_file_file.wav ADDED Viewed

Binary file (271 kB). View file