Generate data visualizations by speaking in your native language ! Try it in Romanian 😉
This demo is wired to the official SD Space and using the Whisperer model
''') gr.Markdown( """ ## 1. Record audio or Upload an audio file: """ ) with gr.Tab(label="Record audio input", elem_id="record_tab"): with gr.Column(): record_input = gr.Audio( source="microphone", type="filepath", show_label=False, elem_id="record_btn" ) with gr.Row(): audio_r_translate = gr.Button("Check Whisper first ? 👍", elem_id="check_btn_1") audio_r_direct_sd = gr.Button("Magic Whisper › SD right now!", elem_id="magic_btn_1") with gr.Tab(label="Upload audio input", elem_id="upload_tab"): with gr.Column(): upload_input = gr.Audio( source="upload", type="filepath", show_label=False, elem_id="upload_area" ) with gr.Row(): audio_u_translate = gr.Button("Check Whisper first ? 👍", elem_id="check_btn_2") # audio_u_direct_sd = gr.Button("Magic Whisper › SD right now!", elem_id="magic_btn_2") with gr.Accordion(label="Stable Diffusion Settings", elem_id="sd_settings", visible=False): with gr.Row(): guidance_scale = gr.Slider(2, 15, value = 7, label = 'Guidance Scale') nb_iterations = gr.Slider(10, 50, value = 25, step = 1, label = 'Steps') seed = gr.Slider(label = "Seed", minimum = 0, maximum = 2147483647, step = 1, randomize = True) gr.Markdown( """ ## 2. Check Whisper output, correct it if necessary: """ ) with gr.Row(): transcripted_output = gr.Textbox( label="Transcription in your detected spoken language", lines=3, elem_id="transcripted" ) language_detected_output = gr.Textbox(label="Native language", elem_id="spoken_lang",lines=3) with gr.Column(): translated_output = gr.Textbox( label="Transcript translated in English by Whisper", lines=4, elem_id="translated" ) with gr.Row(): clear_btn = gr.Button(value="Clear") diffuse_btn = gr.Button(value="OK, Diffuse this prompt !", elem_id="diffuse_btn") codegen_btn = gr.Button(value="Generate code from this prompt!", elem_id="codegen_btn") clear_btn.click(fn=lambda value: gr.update(value=""), inputs=clear_btn, outputs=translated_output) gr.Markdown( """ ## 3. Use whisper results for code generation """ ) with gr.Column(): with gr.Row(): # clear_btn = gr.Button(value="Clear") # diffuse_btn = gr.Button(value="Diffuse this prompt!", elem_id="diffuse_btn") # codegen_btn = gr.Button(value="Generate code from this prompt!", elem_id="codegen_btn") code = gr.Textbox( label="Code generated", lines=4, elem_id="translated" ) # clear_btn.click(fn=lambda value: gr.update(value=""), inputs=clear_btn, outputs=translated_output) # gr.Markdown(""" # ## 4. Plot the generated code ☕️ # Inference time is about ~20-30 seconds, when it's your turn 😬 # """ # ) # with gr.Column(): # with gr.Row(): # def outbreak(plot_type): # df = pd.read_csv('emp_experience_data.csv') # data_encoded = df.copy(deep=True) # categorical_column = ['Attrition', 'Gender', 'BusinessTravel', 'Education', 'EmployeeExperience', 'EmployeeFeedbackSentiments', 'Designation', # 'SalarySatisfaction', 'HealthBenefitsSatisfaction', 'UHGDiscountProgramUsage', 'HealthConscious', 'CareerPathSatisfaction', 'Region'] # label_encoding = LabelEncoder() # for col in categorical_column: # data_encoded[col] = label_encoding.fit_transform(data_encoded[col]) # if plot_type == "Find Data Correlation": # fig = plt.figure() # data_correlation = data_encoded.corr() # sns.heatmap(data_correlation, xticklabels = data_correlation.columns, yticklabels = data_correlation.columns) # return fig # if plot_type == "Filter Correlation Data": # fig = plt.figure() # filtered_df = df[['EmployeeExperience', 'EmployeeFeedbackSentiments', 'Age', 'SalarySatisfaction', 'BusinessTravel', 'HealthBenefitsSatisfaction']] # correlation_filter_data = filtered_df.corr() # sns.heatmap(correlation_filter_data, xticklabels = filtered_df.columns, yticklabels = filtered_df.columns) # return fig # if plot_type == "Age vs Attrition": # fig = plt.figure() # plt.hist(data_encoded['Age'], bins=np.arange(0,80,10), alpha=0.8, rwidth=0.9, color='red') # plt.xlabel("Age") # plt.ylabel("Count") # plt.title("Age vs Attrition") # return fig # if plot_type == "Business Travel vs Attrition": # fig = plt.figure() # ax = sns.countplot(x="BusinessTravel", hue="Attrition", data=data_encoded) # for p in ax.patches: # ax.annotate('{}'.format(p.get_height()), (p.get_x(), p.get_height()+1)) # return fig # if plot_type == "Employee Experience vs Attrition": # fig = plt.figure() # ax = sns.countplot(x="EmployeeExperience", hue="Attrition", data=data_encoded) # for p in ax.patches: # ax.annotate('{}'.format(p.get_height()), (p.get_x(), p.get_height()+1)) # return figure # inputs = [ # gr.Dropdown(["Find Data Correlation", "Filter Correlation Data", "Business Travel vs Attrition", "Employee Experience vs Attrition", "Age vs Attrition",], label="Data Correlation and Visualization") # ] # outputs = gr.Plot() # demo2 = gr.Interface( # fn = outbreak, # inputs = inputs, # outputs = outputs, # title="Employee-Experience: Data Correlation and Pattern Visualization", # allow_flagging=False # ) gr.Markdown(""" ## 5. Wait for Stable Diffusion Results ☕️ Inference time is about ~20-30 seconds, when it's your turn 😬 """ ) sd_output = gr.Gallery().style(grid=2, height="auto") gr.Markdown(""" ### 📌 About the models
Whisper is a general-purpose speech recognition model.
It is trained on a large dataset of diverse audio and is also a multi-task model that can perform multilingual speech recognition as well as speech translation and language identification.
—
Stable Diffusion is a state of the art text-to-image model that generates images from text.
The model is licensed with a CreativeML Open RAIL-M license.
The authors claim no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in this license.
The license forbids you from sharing any content that violates any laws, produce any harm to a person, disseminate any personal information that would be meant for harm, spread misinformation and target vulnerable groups.
For the full list of restrictions please read the license.
Despite how impressive being able to turn text into image is, beware to the fact that this model may output content that reinforces or exacerbates societal biases, as well as realistic faces, pornography and violence.
The model was trained on the LAION-5B dataset, which scraped non-curated image-text-pairs from the internet (the exception being the removal of illegal content) and is meant for research purposes.
You can read more in the model card.