initial

.dockerignore

@@ -0,0 +1,9 @@
+__pycache__/
+*.pyc
+*.pyo
+*.pyd
+venv/
+reports/
+uploads/
+.git/
+.env

.gitignore

@@ -0,0 +1,29 @@
+# Python virtual environment
+venv/
+/venv/
+__pycache__/
+*.pyc
+*.pyo
+*.pyd
+
+# User-specific files
+uploads/
+reports/
+
+# Configuration and secrets
+# IMPORTANT: This prevents your API keys from being uploaded to GitHub
+.env
+
+# IDE and editor files
+.vscode/
+.idea/
+*.swp
+*.swo
+
+# OS-specific files
+.DS_Store
+Thumbs.db
+
+# Matplotlib cache
+*.png
+!/path/to/keep/some/images/

DOCKERFILE

@@ -0,0 +1,25 @@
+# Use an official Python runtime as a parent image
+FROM python:3.11-slim
+
+# Set the working directory in the container
+WORKDIR /app
+
+# Copy the requirements file into the container
+COPY requirements.txt .
+
+# Install any needed system dependencies (like for audio)
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    ffmpeg \
+    && rm -rf /var/lib/apt/lists/*
+
+# Install the Python dependencies
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy the rest of the application's code into the container
+COPY . .
+
+# Expose the port that Gradio runs on
+EXPOSE 7860
+
+# Define the command to run your application
+CMD ["python3", "app.py"]

app.py

@@ -0,0 +1,119 @@
+# app.py
+import gradio as gr
+import os
+import time
+import datetime
+import random
+import config
+from modules.tts_handler import text_to_speech_file
+from modules.stt_handler import transcribe_audio
+from modules.doc_processor import extract_text_from_document
+from modules.llm_handler import generate_question, evaluate_answer
+from modules.report_generator import generate_pdf_report
+
+def start_interview(interview_type, doc_file, name, num_questions):
+    if not interview_type or not doc_file:
+        return {
+            chatbot: gr.update(value=[[None, "Please select an interview type and upload a document to begin."]]),
+            audio_in: gr.update(interactive=False)
+        }
+    doc_text = extract_text_from_document(doc_file.name)
+    if "Error" in doc_text or "Unsupported" in doc_text:
+        return {
+            chatbot: gr.update(value=[[None, f"Error: {doc_text}"]]),
+            audio_in: gr.update(interactive=False)
+        }
+    initial_state = {
+        "interview_type": interview_type, "doc_text": doc_text,
+        "name": name if name else "User", "question_count": int(num_questions),
+        "current_question_num": 1, "interview_log": []
+    }
+    first_question = generate_question(interview_type, doc_text)
+    initial_state["current_question_text"] = first_question
+    greeting = f"Hello {initial_state['name']}. We'll go through {int(num_questions)} questions today. Here is your first question:"
+    tts_prompt = f"{greeting} {first_question}"
+    ai_voice_path = text_to_speech_file(tts_prompt)
+    return {
+        state: initial_state,
+        chatbot: gr.update(value=[[None, f"{greeting}\n\n{first_question}"]]),
+        audio_out: gr.update(value=ai_voice_path, autoplay=True),
+        audio_in: gr.update(interactive=True),
+        start_btn: gr.update(interactive=False)
+    }
+
+def handle_interview_turn(user_audio, chatbot_history, current_state):
+    user_answer_text = transcribe_audio(user_audio)
+    chatbot_history.append([user_answer_text, None])
+    yield {chatbot: chatbot_history, audio_in: gr.update(interactive=False)}
+    
+    evaluation_text = evaluate_answer(current_state["current_question_text"], user_answer_text)
+    current_state["interview_log"].append({
+        "question": current_state["current_question_text"],
+        "answer": user_answer_text,
+        "evaluation": evaluation_text
+    })
+    if current_state["current_question_num"] >= current_state["question_count"]:
+        end_message = "This concludes the interview. Generating your final report now."
+        chatbot_history.append([None, end_message])
+        pdf_path = generate_pdf_file(current_state)
+        ai_voice_path = text_to_speech_file(end_message)
+        yield {
+            chatbot: chatbot_history,
+            audio_out: gr.update(value=ai_voice_path, autoplay=True),
+            download_pdf_btn: gr.update(value=pdf_path, visible=True)
+        }
+    else:
+        current_state["current_question_num"] += 1
+        next_question = generate_question(current_state["interview_type"], current_state["doc_text"])
+        current_state["current_question_text"] = next_question
+        q_num = current_state["current_question_num"]
+        transition_message = f"Thank you. Here is question {q_num}:\n\n{next_question}"
+        chatbot_history.append([None, transition_message])
+        ai_voice_path = text_to_speech_file(transition_message)
+        yield {
+            state: current_state,
+            chatbot: chatbot_history,
+            audio_out: gr.update(value=ai_voice_path, autoplay=True),
+            audio_in: gr.update(interactive=True)
+        }
+
+def generate_pdf_file(state):
+    # --- THIS IS THE CORRECTED LINE ---
+    final_data = { "name": state["name"], "type": state["interview_type"], "q_and_a": state["interview_log"] }
+    file_name = f"Report_{state['name']}_{datetime.datetime.now().strftime('%Y-%m-%d')}.pdf"
+    file_path = os.path.join(config.REPORT_FOLDER, file_name)
+    generate_pdf_report(final_data, file_path)
+    return file_path
+
+with gr.Blocks(theme=gr.themes.Default()) as app:
+    state = gr.State({})
+    gr.Markdown("# 🤖 AI Interview Coach")
+    with gr.Row():
+        with gr.Column(scale=1):
+            gr.Markdown("### Setup")
+            user_name = gr.Textbox(label="Your Name")
+            interview_type_dd = gr.Dropdown(config.INTERVIEW_TYPES, label="Interview Type")
+            num_questions_slider = gr.Slider(minimum=2, maximum=10, value=5, step=1, label="Number of Questions")
+            doc_uploader = gr.File(label="Upload Resume/CV")
+            start_btn = gr.Button("Start Interview", variant="primary")
+        with gr.Column(scale=2):
+            chatbot = gr.Chatbot(label="Conversation", height=500)
+            audio_in = gr.Audio(sources=["microphone"], type="filepath", label="Record Your Answer", interactive=False)
+            download_pdf_btn = gr.File(label="Download Report", visible=False)
+            audio_out = gr.Audio(visible=False, autoplay=True)
+
+    start_btn.click(
+        fn=start_interview,
+        inputs=[interview_type_dd, doc_uploader, user_name, num_questions_slider],
+        outputs=[state, chatbot, audio_out, audio_in, start_btn]
+    )
+    audio_in.stop_recording(
+        fn=handle_interview_turn,
+        inputs=[audio_in, chatbot, state],
+        outputs=[state, chatbot, audio_out, audio_in, download_pdf_btn]
+    )
+
+if __name__ == "__main__":
+    os.makedirs(config.UPLOAD_FOLDER, exist_ok=True)
+    os.makedirs(config.REPORT_FOLDER, exist_ok=True)
+    app.launch(debug=True)

config.py

@@ -0,0 +1,15 @@
+# config.py
+import os
+
+# -- Ollama Configuration --
+OLLAMA_MODEL = 'llama3.1' 
+
+# -- Interview Configuration --
+INTERVIEW_TYPES = ['HR', 'Technical', 'UPSC', 'Group Discussion (GD)', 'Mock IELTS']
+
+# -- Piper TTS Configuration --
+PIPER_VOICE_MODEL = './voice_model/en_US-lessac-medium.onnx' 
+
+# -- Directories --
+UPLOAD_FOLDER = 'uploads'
+REPORT_FOLDER = 'reports'

modules/doc_processor.py

@@ -0,0 +1,29 @@
+# modules/doc_processor.py
+
+import fitz  # PyMuPDF for PDFs
+import docx  # python-docx for DOCX files
+import os
+
+def extract_text_from_document(file_path):
+    """
+    Extracts text from a given document (PDF or DOCX).
+    """
+    text = ""
+    try:
+        _, file_extension = os.path.splitext(file_path)
+
+        if file_extension.lower() == '.pdf':
+            with fitz.open(file_path) as doc:
+                for page in doc:
+                    text += page.get_text()
+        elif file_extension.lower() == '.docx':
+            doc = docx.Document(file_path)
+            for para in doc.paragraphs:
+                text += para.text + "\n"
+        else:
+            return "Unsupported file format. Please upload a .pdf or .docx file."
+    
+    except Exception as e:
+        return f"Error reading document: {e}"
+
+    return text

modules/llm_handler.py

@@ -0,0 +1,36 @@
+# modules/llm_handler.py
+import os
+from groq import Groq
+from modules.web_search import search_for_example_answers
+
+# Initialize the Groq client
+# The API key will be automatically read from the HF Space's secrets
+client = Groq(api_key=os.environ.get("GROQ_API_KEY"))
+
+def generate_question(interview_type, document_text):
+    prompt = f"As an expert {interview_type} interviewer, ask one relevant question based on this document: --- {document_text} ---"
+    try:
+        chat_completion = client.chat.completions.create(
+            messages=[{"role": "user", "content": prompt}],
+            model="llama3-8b-8192",
+        )
+        return chat_completion.choices[0].message.content
+    except Exception as e:
+        return f"Error generating question: {e}"
+
+def evaluate_answer(question, answer):
+    example_answers = search_for_example_answers(question)
+    prompt = f"""
+    You are an interview coach. Compare the candidate's answer to the expert examples and provide a concise evaluation.
+    Question: "{question}"
+    Candidate's Answer: "{answer}"
+    Expert Examples: --- {example_answers} ---
+    """
+    try:
+        chat_completion = client.chat.completions.create(
+            messages=[{"role": "user", "content": prompt}],
+            model="llama3-8b-8192",
+        )
+        return chat_completion.choices[0].message.content
+    except Exception as e:
+        return f"An error occurred during evaluation: {e}"

modules/report_generator.py

@@ -0,0 +1,114 @@
+# modules/report_generator.py
+
+import datetime
+import os
+import numpy as np
+import matplotlib.pyplot as plt
+from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer, PageBreak, Image, Frame, PageTemplate
+from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
+from reportlab.lib.enums import TA_JUSTIFY, TA_CENTER, TA_LEFT
+from reportlab.lib.units import inch
+from reportlab.lib import colors
+from modules.llm_handler import generate_holistic_feedback, parse_scores_from_evaluation
+import config
+
+# --- Page Template with Header and Footer (Unchanged) ---
+class ReportPageTemplate(PageTemplate):
+    def __init__(self, id, pagesize):
+        frame = Frame(inch, inch, pagesize[0] - 2 * inch, pagesize[1] - 2 * inch, id='normal')
+        PageTemplate.__init__(self, id, [frame])
+
+    def beforeDrawPage(self, canvas, doc):
+        canvas.saveState()
+        canvas.setFont('Helvetica', 9)
+        canvas.setFillColor(colors.grey)
+        footer_text = f"Page {doc.page} | AI Interview Coach Report | Generated on {datetime.datetime.now().strftime('%Y-%m-%d')}"
+        canvas.drawCentredString(doc.width / 2 + inch, 0.75 * inch, footer_text)
+        canvas.restoreState()
+
+def create_radar_chart(labels, scores, file_path):
+    # This function is unchanged
+    num_vars = len(labels)
+    angles = np.linspace(0, 2 * np.pi, num_vars, endpoint=False).tolist()
+    scores += scores[:1]
+    angles += angles[:1]
+    fig, ax = plt.subplots(figsize=(6, 6), subplot_kw=dict(polar=True))
+    ax.fill(angles, scores, color='#4A90E2', alpha=0.2)
+    ax.plot(angles, scores, color='#4A90E2', linewidth=2, linestyle='solid')
+    ax.set_yticklabels([])
+    ax.set_xticks(angles[:-1])
+    ax.set_xticklabels(labels, size=12, color='grey')
+    ax.set_rlabel_position(30)
+    ax.set_ylim(0, 10)
+    for angle, score in zip(angles[:-1], scores[:-1]):
+        ax.text(angle, score + 1.5, str(score), ha='center', va='center', size=14, color="#000000", weight='bold')
+    plt.title('Performance Snapshot', size=20, color='#333333', y=1.1)
+    os.makedirs(os.path.dirname(file_path), exist_ok=True)
+    plt.savefig(file_path, transparent=True, dpi=150)
+    plt.close(fig)
+    print(f"📈 Radar chart saved to {file_path}")
+
+def generate_pdf_report(interview_data, file_path):
+    doc = SimpleDocTemplate(file_path, pagesize=(8.5 * inch, 11 * inch),
+                            leftMargin=inch, rightMargin=inch, topMargin=inch, bottomMargin=inch)
+    doc.addPageTemplates([ReportPageTemplate('main_template', (8.5 * inch, 11 * inch))])
+    
+    styles = getSampleStyleSheet()
+    
+    # --- THIS IS THE CORRECTED SECTION ---
+    # Instead of adding styles with existing names, we create new ones with unique names.
+    styles.add(ParagraphStyle(name='ReportTitle', parent=styles['h1'], fontSize=28, alignment=TA_CENTER, spaceAfter=24))
+    styles.add(ParagraphStyle(name='ReportSubTitle', parent=styles['h2'], fontSize=16, alignment=TA_CENTER, spaceAfter=12, textColor=colors.HexColor('#555555')))
+    styles.add(ParagraphStyle(name='Justify', alignment=TA_JUSTIFY, spaceAfter=12, leading=14))
+    styles.add(ParagraphStyle(name='MainHeader', parent=styles['h1'], fontSize=22, spaceBefore=12, spaceAfter=20, alignment=TA_LEFT, textColor=colors.HexColor('#2c3e50')))
+    styles.add(ParagraphStyle(name='QuestionTitle', parent=styles['h2'], spaceBefore=20, spaceAfter=10, textColor=colors.HexColor('#2980b9')))
+    styles.add(ParagraphStyle(name='SectionTitle', parent=styles['h3'], spaceBefore=12, spaceAfter=6, textColor=colors.HexColor('#34495e')))
+    # --- END OF CORRECTION ---
+
+    story = []
+
+    # --- 1. The Title Page ---
+    story.append(Paragraph("Interview Performance Report", styles['ReportTitle']))
+    story.append(Spacer(1, 0.5 * inch))
+    story.append(Paragraph(f"Prepared for: <b>{interview_data.get('name', 'N/A')}</b>", styles['ReportSubTitle']))
+    story.append(Spacer(1, 0.2 * inch))
+    story.append(Paragraph(f"Interview Type: <b>{interview_data['type']}</b>", styles['ReportSubTitle']))
+    story.append(Spacer(1, 0.2 * inch))
+    story.append(Paragraph(f"Date of Report: <b>{datetime.datetime.now().strftime('%B %d, %Y')}</b>", styles['ReportSubTitle']))
+    story.append(PageBreak())
+
+    # --- 2. Overall Performance & Graph Section ---
+    story.append(Paragraph("Overall Performance Analysis", styles['MainHeader']))
+    full_log_text = ""
+    all_scores = []
+    skill_labels = ['Factual Accuracy', 'Relevance & Directness', 'Structure & Clarity']
+    for i, qa in enumerate(interview_data['q_and_a']):
+        full_log_text += f"Q{i+1}: {qa['question']}\nA: {qa['answer']}\n---\n"
+        scores = parse_scores_from_evaluation(qa['evaluation'])
+        all_scores.append([scores.get(label, 0) for label in skill_labels])
+    holistic_feedback = generate_holistic_feedback(full_log_text).replace('\n', '<br/>')
+    story.append(Paragraph(holistic_feedback, styles['Justify']))
+    story.append(Spacer(1, 0.3 * inch))
+
+    if all_scores:
+        avg_scores = np.mean(all_scores, axis=0).tolist()
+        chart_path = os.path.join(config.REPORT_FOLDER, "skill_chart.png")
+        if os.path.exists(chart_path): os.remove(chart_path)
+        create_radar_chart(skill_labels, avg_scores, chart_path)
+        story.append(Image(chart_path, width=4.5*inch, height=4.5*inch, hAlign='CENTER'))
+    story.append(PageBreak())
+
+    # --- 3. Detailed Question-by-Question Analysis ---
+    story.append(Paragraph("Detailed Question Analysis", styles['MainHeader']))
+    for i, qa in enumerate(interview_data['q_and_a']):
+        story.append(Paragraph(f"Question {i+1}: {qa['question']}", styles['QuestionTitle']))
+        story.append(Paragraph("Your Answer:", styles['SectionTitle']))
+        story.append(Paragraph(qa.get('answer', 'N/A'), styles['Justify']))
+        story.append(Paragraph("AI Evaluation:", styles['SectionTitle']))
+        story.append(Paragraph(qa.get('evaluation', 'N/A').replace('\n', '<br/>'), styles['Justify']))
+
+    try:
+        doc.build(story)
+        print(f"\n✅ Professional report generated successfully: {file_path}")
+    except Exception as e:
+        print(f"💥 Error generating professional PDF report: {e}")

modules/stt_handler.py

@@ -0,0 +1,32 @@
+# modules/stt_handler.py
+import speech_recognition as sr
+import os
+
+def transcribe_audio(audio_filepath):
+    if not audio_filepath or not os.path.exists(audio_filepath):
+        print("STT Error: No audio file provided or file does not exist.")
+        return "[Transcription error: Invalid audio file path]"
+    
+    recognizer = sr.Recognizer()
+    try:
+        with sr.AudioFile(audio_filepath) as source:
+            audio_data = recognizer.record(source)
+            print("Transcribing with Whisper...")
+            text = recognizer.recognize_whisper(audio_data, language="english")
+            print(f"User transcribed as: {text}")
+            return text
+    except sr.UnknownValueError:
+        print("STT Error: Whisper could not understand the audio.")
+        return "[Could not understand audio]"
+    except sr.RequestError as e:
+        print(f"STT Error: Could not request results from Whisper service; {e}")
+        return f"[Transcription error: Whisper service issue - {e}]"
+    except Exception as e:
+        print(f"STT Error: An unexpected error occurred during transcription: {e}")
+        return f"[Transcription error: {e}]"
+    finally:
+        if os.path.exists(audio_filepath):
+            try:
+                os.remove(audio_filepath)
+            except OSError as e:
+                print(f"Error deleting temp audio file {audio_filepath}: {e}")

modules/tts_handler.py

@@ -0,0 +1,31 @@
+# modules/tts_handler.py
+import subprocess
+import platform
+import sys
+import os
+import config
+from pydub import AudioSegment
+import tempfile
+
+def text_to_speech_file(text_to_speak):
+    print(f"AI generating audio for: {text_to_speak}")
+    piper_executable = 'piper'  # Use system PATH
+    try:
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".raw") as raw_file:
+            raw_filename = raw_file.name
+
+        command = [piper_executable, '--model', config.PIPER_VOICE_MODEL, '--output-file', raw_filename]
+        process = subprocess.Popen(command, stdin=subprocess.PIPE)
+        process.communicate(input=text_to_speak.encode('utf-8'))
+
+        raw_audio = AudioSegment.from_file(raw_filename, format="raw", frame_rate=22050, channels=1, sample_width=2)
+
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as wav_file:
+            wav_filename = wav_file.name
+
+        raw_audio.export(wav_filename, format="wav")
+        os.remove(raw_filename)
+        return wav_filename
+    except Exception as e:
+        print(f"An error occurred during TTS generation: {e}")
+        return None

modules/web_search.py

@@ -0,0 +1,30 @@
+# modules/web_search.py
+
+from ddgs import DDGS
+
+def search_for_example_answers(query: str, num_results: int = 2):
+    """
+    Performs a targeted web search for high-quality example answers to an interview question.
+    """
+    # Refine the query to find expert answers
+    search_query = f"expert sample answer for interview question: \"{query}\""
+    print(f"🌐 Searching for expert answers with query: '{search_query}'")
+    
+    try:
+        with DDGS(timeout=10) as ddgs:
+            results = list(ddgs.text(search_query, max_results=num_results))
+            
+            if not results:
+                print("   -> No example answers found.")
+                return "No example answers found on the web."
+            
+            formatted_results = ""
+            for i, res in enumerate(results):
+                formatted_results += f"Example Answer Source {i+1}:\nTitle: {res.get('title', 'N/A')}\nSnippet: {res.get('body', 'N/A')}\n\n"
+            
+            print(f"   -> Found {len(results)} example answers.")
+            return formatted_results
+            
+    except Exception as e:
+        print(f"💥 Web search failed: {e}")
+        return "Web search for example answers failed."

requirements.txt

@@ -0,0 +1,16 @@
+ollama
+openai-whisper
+gradio
+pydub
+soundfile
+pyaudio
+piper-tts
+PyMuPDF
+python-docx
+reportlab
+speechrecognition
+duckduckgo-search
+ddgs
+matplotlib
+regex
+groq

voice_model/en_US-lessac-medium.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5efe09e69902187827af646e1a6e9d269dee769f9877d17b16b1b46eeaaf019f
+size 63201294

voice_model/en_US-lessac-medium.onnx.json

@@ -0,0 +1,493 @@
+{
+  "audio": {
+    "sample_rate": 22050,
+    "quality": "medium"
+  },
+  "espeak": {
+    "voice": "en-us"
+  },
+  "inference": {
+    "noise_scale": 0.667,
+    "length_scale": 1,
+    "noise_w": 0.8
+  },
+  "phoneme_type": "espeak",
+  "phoneme_map": {},
+  "phoneme_id_map": {
+    "_": [
+      0
+    ],
+    "^": [
+      1
+    ],
+    "$": [
+      2
+    ],
+    " ": [
+      3
+    ],
+    "!": [
+      4
+    ],
+    "'": [
+      5
+    ],
+    "(": [
+      6
+    ],
+    ")": [
+      7
+    ],
+    ",": [
+      8
+    ],
+    "-": [
+      9
+    ],
+    ".": [
+      10
+    ],
+    ":": [
+      11
+    ],
+    ";": [
+      12
+    ],
+    "?": [
+      13
+    ],
+    "a": [
+      14
+    ],
+    "b": [
+      15
+    ],
+    "c": [
+      16
+    ],
+    "d": [
+      17
+    ],
+    "e": [
+      18
+    ],
+    "f": [
+      19
+    ],
+    "h": [
+      20
+    ],
+    "i": [
+      21
+    ],
+    "j": [
+      22
+    ],
+    "k": [
+      23
+    ],
+    "l": [
+      24
+    ],
+    "m": [
+      25
+    ],
+    "n": [
+      26
+    ],
+    "o": [
+      27
+    ],
+    "p": [
+      28
+    ],
+    "q": [
+      29
+    ],
+    "r": [
+      30
+    ],
+    "s": [
+      31
+    ],
+    "t": [
+      32
+    ],
+    "u": [
+      33
+    ],
+    "v": [
+      34
+    ],
+    "w": [
+      35
+    ],
+    "x": [
+      36
+    ],
+    "y": [
+      37
+    ],
+    "z": [
+      38
+    ],
+    "æ": [
+      39
+    ],
+    "ç": [
+      40
+    ],
+    "ð": [
+      41
+    ],
+    "ø": [
+      42
+    ],
+    "ħ": [
+      43
+    ],
+    "ŋ": [
+      44
+    ],
+    "œ": [
+      45
+    ],
+    "ǀ": [
+      46
+    ],
+    "ǁ": [
+      47
+    ],
+    "ǂ": [
+      48
+    ],
+    "ǃ": [
+      49
+    ],
+    "ɐ": [
+      50
+    ],
+    "ɑ": [
+      51
+    ],
+    "ɒ": [
+      52
+    ],
+    "ɓ": [
+      53
+    ],
+    "ɔ": [
+      54
+    ],
+    "ɕ": [
+      55
+    ],
+    "ɖ": [
+      56
+    ],
+    "ɗ": [
+      57
+    ],
+    "ɘ": [
+      58
+    ],
+    "ə": [
+      59
+    ],
+    "ɚ": [
+      60
+    ],
+    "ɛ": [
+      61
+    ],
+    "ɜ": [
+      62
+    ],
+    "ɞ": [
+      63
+    ],
+    "ɟ": [
+      64
+    ],
+    "ɠ": [
+      65
+    ],
+    "ɡ": [
+      66
+    ],
+    "ɢ": [
+      67
+    ],
+    "ɣ": [
+      68
+    ],
+    "ɤ": [
+      69
+    ],
+    "ɥ": [
+      70
+    ],
+    "ɦ": [
+      71
+    ],
+    "ɧ": [
+      72
+    ],
+    "ɨ": [
+      73
+    ],
+    "ɪ": [
+      74
+    ],
+    "ɫ": [
+      75
+    ],
+    "ɬ": [
+      76
+    ],
+    "ɭ": [
+      77
+    ],
+    "ɮ": [
+      78
+    ],
+    "ɯ": [
+      79
+    ],
+    "ɰ": [
+      80
+    ],
+    "ɱ": [
+      81
+    ],
+    "ɲ": [
+      82
+    ],
+    "ɳ": [
+      83
+    ],
+    "ɴ": [
+      84
+    ],
+    "ɵ": [
+      85
+    ],
+    "ɶ": [
+      86
+    ],
+    "ɸ": [
+      87
+    ],
+    "ɹ": [
+      88
+    ],
+    "ɺ": [
+      89
+    ],
+    "ɻ": [
+      90
+    ],
+    "ɽ": [
+      91
+    ],
+    "ɾ": [
+      92
+    ],
+    "ʀ": [
+      93
+    ],
+    "ʁ": [
+      94
+    ],
+    "ʂ": [
+      95
+    ],
+    "ʃ": [
+      96
+    ],
+    "ʄ": [
+      97
+    ],
+    "ʈ": [
+      98
+    ],
+    "ʉ": [
+      99
+    ],
+    "ʊ": [
+      100
+    ],
+    "ʋ": [
+      101
+    ],
+    "ʌ": [
+      102
+    ],
+    "ʍ": [
+      103
+    ],
+    "ʎ": [
+      104
+    ],
+    "ʏ": [
+      105
+    ],
+    "ʐ": [
+      106
+    ],
+    "ʑ": [
+      107
+    ],
+    "ʒ": [
+      108
+    ],
+    "ʔ": [
+      109
+    ],
+    "ʕ": [
+      110
+    ],
+    "ʘ": [
+      111
+    ],
+    "ʙ": [
+      112
+    ],
+    "ʛ": [
+      113
+    ],
+    "ʜ": [
+      114
+    ],
+    "ʝ": [
+      115
+    ],
+    "ʟ": [
+      116
+    ],
+    "ʡ": [
+      117
+    ],
+    "ʢ": [
+      118
+    ],
+    "ʲ": [
+      119
+    ],
+    "ˈ": [
+      120
+    ],
+    "ˌ": [
+      121
+    ],
+    "ː": [
+      122
+    ],
+    "ˑ": [
+      123
+    ],
+    "˞": [
+      124
+    ],
+    "β": [
+      125
+    ],
+    "θ": [
+      126
+    ],
+    "χ": [
+      127
+    ],
+    "ᵻ": [
+      128
+    ],
+    "ⱱ": [
+      129
+    ],
+    "0": [
+      130
+    ],
+    "1": [
+      131
+    ],
+    "2": [
+      132
+    ],
+    "3": [
+      133
+    ],
+    "4": [
+      134
+    ],
+    "5": [
+      135
+    ],
+    "6": [
+      136
+    ],
+    "7": [
+      137
+    ],
+    "8": [
+      138
+    ],
+    "9": [
+      139
+    ],
+    "̧": [
+      140
+    ],
+    "̃": [
+      141
+    ],
+    "̪": [
+      142
+    ],
+    "̯": [
+      143
+    ],
+    "̩": [
+      144
+    ],
+    "ʰ": [
+      145
+    ],
+    "ˤ": [
+      146
+    ],
+    "ε": [
+      147
+    ],
+    "↓": [
+      148
+    ],
+    "#": [
+      149
+    ],
+    "\"": [
+      150
+    ],
+    "↑": [
+      151
+    ],
+    "̺": [
+      152
+    ],
+    "̻": [
+      153
+    ]
+  },
+  "num_symbols": 256,
+  "num_speakers": 1,
+  "speaker_id_map": {},
+  "piper_version": "1.0.0",
+  "language": {
+    "code": "en_US",
+    "family": "en",
+    "region": "US",
+    "name_native": "English",
+    "name_english": "English",
+    "country_english": "United States"
+  },
+  "dataset": "lessac"
+}