🎙 AssemblyAI Tk Transcriber (30 min chunks + overlap)

A small desktop app (Tkinter) to transcribe local audio files with AssemblyAI.

It:

• Loads .m4a, .mp3, .wav, etc. (via ffmpeg + pydub)
• Lets you choose the language (or auto-detect)
• Splits audio into 60s chunks with overlap (configurable)
• Saves per-chunk transcripts and a full joined transcript

⚠️ You need an AssemblyAI API key (paid or free tier). Don’t commit it—store it in .env.

---

✨ Features

• Simple GUI (pick file, language, chunk size, overlap)
• Works with .m4a (common from iPhone voice notes) and many formats
• Safer on Python 3.13 with audioop-lts backport
• Outputs:
  • YourFile_FULL.txt
  • YourFile_parts/YourFile_partXXX_...txt

---

🖥 Windows Installation (Step-by-Step)

1️⃣ Install Chocolatey (One-time setup)

Chocolatey makes it easy to install ffmpeg.

1. Open PowerShell as Administrator

   • Press Start, type PowerShell, right-click, select Run as administrator.

2. Run this command:

   Set-ExecutionPolicy Bypass -Scope Process -Force; [System.Net.ServicePointManager]::SecurityProtocol = [System.Net.ServicePointManager]::SecurityProtocol -bor 3072; iex ((New-Object System.Net.WebClient).DownloadString('https://community.chocolatey.org/install.ps1'))
   

3. Close and reopen PowerShell to refresh PATH. Verify:

   choco -v

---

2️⃣ Install ffmpeg

In Administrator PowerShell:

choco install ffmpeg -y

Verify:

ffmpeg -version

If you see version info, ffmpeg is ready.

---

3️⃣ Clone and prepare the project

1. Open a normal PowerShell (no admin) and go to your folder:

   cd "C:\Path\To\Transcribe"

2. Create and activate a virtual environment:

   python -m venv .venv
   .\.venv\Scripts\Activate.ps1

3. Install dependencies:

   pip install -U pip
   pip install assemblyai pydub python-dotenv audioop-lts

---

4️⃣ Add your API key

Create a file .env in the project root:

ASSEMBLYAI_API_KEY=your_real_api_key_here

Or paste it directly in the app when running.

---

5️⃣ Run the app

python main.py

• Pick your audio file
• Choose a language or ALD for auto-detect
• Adjust chunk/overlap if needed
• Click Transcribe

---

📂 Output

• Full transcript: YourFile_FULL.txt
• Per-chunk transcripts: YourFile_parts/*.txt with timestamps in filenames

---

🛠 Troubleshooting

“ffmpeg is not recognized”

• Make sure ffmpeg -version works in PowerShell

• If not, reinstall via Chocolatey (admin):

  choco install ffmpeg -y

ModuleNotFoundError: audioop on Python 3.13

• Install backport:

  pip install audioop-lts

• Or use Python 3.12 for maximum compatibility

Large files or rate limits

• Keep chunk size at 60s
• Use 3–5s overlap for context continuity

---

🔒 Security & Privacy

• API key stays local (in .env or entered in GUI)
• Audio and transcripts are ignored by .gitignore to avoid accidental commits