A Sound category, and a Sound tab that proves your speakers work

Making a noise from Python meant installing something, and the usual
first choice, pyttsx3, says nothing at all until runAndWait is called.
So the honest answer to "why can I hear nothing" was rarely obvious.

Every computer already has a voice and a speaker; the trouble is that
each one is reached a different way. Windows has the voices behind
Narrator, macOS has say, Linux has espeak. The new Sound category hides
that difference behind twenty two blocks that work everywhere with
nothing installed: say out loud, start saying, set voice, set speaking
speed, save speech to a file, play note, play drum, play a frequency in
any of five wave shapes, tempo, rest, play and start a sound file, stop,
volume.

Tones, notes and drums are worked out one sample at a time and written
as real wav files, so there is no synthesiser to install either. Each
one is calculated once and kept, so a note inside a loop stays fast. The
voice dropdown lists the voices this computer really has, looked up in
the background so the window never waits for it.

The Sound tab on the right is where you go when nothing can be heard.
Test my speakers plays three notes and says so; if you hear nothing, the
mute switch or the output device is the culprit rather than your program.
Type a sentence, choose a voice, drag the speed and volume, press Say it.
Put these blocks in my project then drops blocks that do exactly what you
just heard. It runs the same helper functions the blocks compile to, so
the tab and your program can never drift apart.

For anyone who would rather use pyttsx3, gtts or playsound there are now
hand written blocks for those too, and the pyttsx3 ones always call
runAndWait.

Also here: the side strip gives its twelfth category room, the panel tabs
no longer clip, and the block drawer starts a little wider.

Version 1.5.0.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Author: ZDStudios

README.md

@@ -8,7 +8,7 @@
 
 [![Python](https://img.shields.io/badge/python-3.9%2B-4C97FF?logo=python&logoColor=white)](https://www.python.org/)
 [![Dependencies](https://img.shields.io/badge/dependencies-none-59C059)](#)
-[![One file](https://img.shields.io/badge/one%20file-7%2C000%20lines-FFAB19)](scratchpy_studio.py)
+[![One file](https://img.shields.io/badge/one%20file-9%2C700%20lines-FFAB19)](scratchpy_studio.py)
 [![Build](https://github.com/ZDStudios/scratchpy-studio/actions/workflows/build-apps.yml/badge.svg)](https://github.com/ZDStudios/scratchpy-studio/actions/workflows/build-apps.yml)
 [![Download](https://img.shields.io/badge/download-windows%20%7C%20macos%20%7C%20linux-855CD6)](https://github.com/ZDStudios/scratchpy-studio/releases/latest)
 [![License](https://img.shields.io/badge/license-MIT-0FBD8C)](LICENSE)
@@ -49,11 +49,12 @@ and hand in.
 
 |  | |
 |---|---|
-| 🧩 **148 blocks** | Hat blocks, C-shaped loops, hexagonal booleans, reporter ovals that drop into slots — the real Scratch 3 shapes and colours |
+| 🧩 **177 blocks** | Hat blocks, C-shaped loops, hexagonal booleans, reporter ovals that drop into slots — the real Scratch 3 shapes and colours |
 | 🐍 **Real Python, live** | The generated source updates as you drag. No hidden interpreter |
 | ▶️ **It actually runs** | `print`, `input`, errors and a stop button, all wired to the built-in console |
 | 👆 **Click a block to try it** | A loose block runs on its own and reports what it printed in a bubble underneath |
 | 🌐 **Talks to the web out of the box** | GET, POST, headers, JSON, downloads — using `urllib` from the standard library, so there is nothing to install |
+| 🔊 **It talks** | A Sound category that speaks, plays notes and beats drums — using the voice and speaker the computer already has |
 | 📥 **Import any `.py`** | Turn a program you already have into blocks — loops, functions, try/except, f-strings and all |
 | 📦 **Every PyPI package** | pip dashboard installs anything and turns it into blocks automatically |
 | 🤖 **Works with AI** | Built-in MCP server so an assistant can build blocks alongside you |
@@ -147,6 +148,56 @@ test is exactly what your program will do.
 
 ---
 
+## It talks. Nothing installed.
+
+Every computer already has a voice and a speaker — the trouble is that each one
+reaches them a different way. Windows has the voices behind Narrator, macOS has
+`say`, Linux has `espeak`. The **Sound** category hides the difference, so one
+block works on all three.
+
+<div align="center">
+<img src="docs/screenshot-sound.png" width="900" alt="The Sound category and the Sound tab">
+</div>
+
+| Block | What it does |
+|---|---|
+| `say [hello] out loud` | Speaks, and waits until the sentence is finished |
+| `start saying [hello]` | Speaks while the program carries on |
+| `set voice to (▾)` | The dropdown lists the voices *this* computer really has |
+| `set speaking speed to (0)` | −10 is very slow, 10 is very fast |
+| `save speech [hello] to [speech.wav]` | Keeps it as a real sound file |
+| `play note (C4▾) for (0.5) beats` | Middle C is `C4` or `60`; `F#3` and `Bb4` work too |
+| `play drum (snare▾) for (0.25) beats` | Twelve drums, synthesised from scratch |
+| `play (440) Hz for (0.5) seconds` | Any frequency, plus sine, square, saw, triangle and noise |
+| `set tempo to (120) beats a minute` · `rest for (0.25) beats` | Music timing, the way Scratch does it |
+| `play sound [sound.wav]` · `start sound [sound.wav]` | Your own files — `.wav` everywhere, `.mp3` on Windows and macOS |
+| `stop all sounds` · `set volume to (100) percent` | Silence, and loudness |
+
+Tones, notes and drums are worked out one sample at a time and written as real
+`.wav` files, so there is no synthesiser to install either. Each one is
+calculated once and then reused, so a note inside a loop stays fast.
+
+### Can you hear this?
+
+The **Sound** tab on the right is a place to make a noise *now*, before you have
+built anything:
+
+- **Test my speakers** plays three notes. If you hear nothing, the problem is
+  the mute switch or the output device, not your program — and it says so.
+- Type a sentence, pick a voice, drag the speed and volume, press **Say it**.
+- A little keyboard for trying notes and a drum.
+- **Put these blocks in my project** drops blocks that do exactly what you just
+  heard, sliders and all.
+
+It runs the *same* helper functions your blocks compile to, so what you hear in
+the tab is what your program will do.
+
+> Using `pyttsx3` instead? Its catch is that nothing is heard until
+> `runAndWait()` is called. ScratchPy now ships hand-written blocks for it that
+> always do both — as it does for `gtts` and `playsound`.
+
+---
+
 ## Bring your own Python
 
 Press **Import .py** and pick any file. The whole program becomes blocks.
@@ -357,7 +408,7 @@ writes a tidy module with a `main()` and an `if __name__ == "__main__":` guard.
 python scratchpy_studio.py --selftest
 ```
 
-Builds the entire interface head-lessly, then **compiles every one of the 148
+Builds the entire interface head-lessly, then **compiles every one of the 177
 blocks**, checks the block definitions are well formed, round-trips a project
 through save and load, exercises the Python importer and the MCP server, and
 finally runs the generated example program and checks its output.

build_apps.py

@@ -74,7 +74,10 @@ def build():
             "--add-data", ASSETS + os.pathsep + "scratchpy_assets"]
 
     if system == "Windows":
-        args += ["--icon", os.path.join(ASSETS, "scratchpy_icon.ico")]
+        args += ["--icon", os.path.join(ASSETS, "scratchpy_icon.ico"),
+                 # the sound blocks reach for winsound from inside a string of
+                 # generated code, so PyInstaller cannot see it by itself
+                 "--hidden-import", "winsound"]
     elif system == "Darwin":
         args += ["--icon", os.path.join(ASSETS, "scratchpy_icon.icns"),
                  "--osx-bundle-identifier", "com.scratchpy.studio"]