SpeakEasy Studio

SpeakEasy Studio is a Windows-first desktop application for text-to-speech and audio-to-text workflows. It combines a modern CustomTkinter interface with multiple synthesis engines, optional summarization, transcript processing, and persistent history/configuration.

Application Screenshots

Text To Speech Tab

Audio To Text Tab

Settings Tab

What This Project Does

• Converts text into speech audio with selectable voices and output formats.
• Transcribes audio files to text using Whisper.
• Applies optional readability enhancements before synthesis.
• Supports summary generation before conversion.
• Tracks output history and provides built-in playback controls.

Core Features

Text To Speech

• Input sources: paste text or load TXT, MD, PDF, DOCX files.
• PDF page-range support.
• Optional summarization in the input tab:
  • sumy for lightweight extraction.
  • bart for higher-quality abstractive summaries.
• Readability enhancements with inline controls:
  • Pause enhancement level (off, mild, strong)
  • Newline normalization
  • Heuristic punctuation insertion
  • List pause enhancement
  • Paragraph pause enhancement
  • Edge fallback pause behavior
• Playback highlighting in the editor during audio playback.
• Dedicated Arabic Editor button (RTL mode) that opens a Qt-based editor for robust Arabic typing, wrapping, and selection.

Speech To Text

• Audio transcription via Whisper (tiny, base, small).
• Optional transcript cleaning and technical normalization.
• Optional emotion/sentiment analysis pipeline and synthesis hints.
• Output as plain text or markdown.
• Send transcript directly to Text to Speech tab.

TTS Engines

• Edge TTS (online neural voices).
• Piper TTS (offline local models).

Voice And Playback

• Voice browser and refresh.
• Piper catalog search for undiscovered voices from multiple sources.
• Source toggles for Piper catalog providers (Hugging Face and project catalog), persisted across app restarts.
• Undownloaded Piper voices are marked visually in the voice list.
• Selecting an undownloaded Piper voice prompts for download.
• Download progress shows percentage, downloaded/total size, transfer speed, and ETA.
• Favorites and per-engine last voice memory.
• Rate, pitch, and volume control.
• Output formats: MP3 and WAV.
• Built-in player bar:
  • Play/pause, stop, seek, speed control, live time updates.

History And Persistence

• Conversion history in src/output/history.json.
• Config persistence in src/config.json.
• Theme and processing preferences saved across sessions.

Tech Stack

• Language: Python 3.x
• UI: CustomTkinter + ttk
• TTS: edge-tts, piper-tts
• STT: openai-whisper
• Summarization: sumy, transformers, torch
• Document parsing: pdfplumber, python-docx, markdown-it-py
• External RTL editor: PySide6
• Media processing/playback: ffmpeg/ffplay (winsound fallback for limited playback)

Project Structure

.
|-- justfile
|-- plan.md
|-- src/
|   |-- main.py
|   |-- config.json
|   |-- requirements.txt
|   |-- core/
|   |-- ui/
|   |-- models/
|   `-- output/
|-- docs/
|   |-- FEATURES.md
|   |-- USAGE.md
|   |-- JUSTFILE.md
|   `-- ARCHITECTURE.md
`-- README.md

Getting Started

Prerequisites

• Windows PowerShell
• Python 3.10+
• ffmpeg available in PATH
• Optional: just command runner
• PySide6 (installed via requirements) for the external Arabic editor window

Quick Start With just

just venv
just install
just run
just run mode="new"

Quick Start Without just

python -m venv .venv
.\.venv\Scripts\python.exe -m pip install -r .\src\requirements.txt
.\.venv\Scripts\python.exe .\src\main.py

justfile Overview

The project ships with an automation file for local development, verification, and packaging.

Key commands:

• just help
• just install
• just run (legacy default)
• just run mode="new" (PySide6 migration UI)
• just run-old
• just run-new
• just compile
• just smoke
• just smoke-old
• just smoke-new
• just stability-startup-cycles cycles="3"
• just stability-long-sessions
• just stability-cancel-recovery
• just phase5-automated
• just verify-tts
• just piper-list
• just piper-download-default
• just build
• just clean

Full reference: see docs/JUSTFILE.md.

Keyboard Shortcuts

• Ctrl+O: open file
• Ctrl+Enter: start conversion
• Ctrl+S: start conversion
• Ctrl+Shift+V: focus text input tab/editor
• Space: play/pause (outside text-input widgets)
• Editor-focused shortcuts (IME/language-independent path):
  • Ctrl+C, Ctrl+V, Ctrl+X, Ctrl+A
  • Ctrl+Z, Ctrl+Y, Ctrl+Shift+Z

Documentation Index

• Feature reference: docs/FEATURES.md
• Usage guide: docs/USAGE.md
• Architecture and flow: docs/ARCHITECTURE.md
• just commands reference: docs/JUSTFILE.md

Notes

• First run for some models (Whisper/BART/Piper voices) may require downloads.
• Long-running operations execute in background threads with cancellation support.
• This project is currently optimized for Windows workflows.