domi-v3-human.yml

name: Domi v3 Human Voice Test

on:
  workflow_dispatch:

jobs:
  generate:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-python@v6.2.0
        with:
          python-version: "3.11"
      - name: Generate Domi v3 samples with audio tags
        env:
          ELEVENLABS_API_KEY: ${{ secrets.ELEVENLABS_API_KEY }}
        run: |
          python3 << 'PYEOF'
          import json, os, urllib.request

          API = "https://api.elevenlabs.io"
          KEY = os.environ["ELEVENLABS_API_KEY"]
          DOMI_ID = "AZnzlk1XvdvUeBnXmlld"

          # Research-backed settings for human-sounding short commands
          SETTINGS = {"stability": 0.45, "similarity_boost": 0.73, "style": 0, "use_speaker_boost": True}

          # v3 with audio tags for commanding delivery
          PHRASES = {
              "move_intense": "[intense] Move... with a PURPOSE!",
              "no_hesitation_commanding": "[commanding] No hesitation. Move!",
              "stay_shouting": "[shouting] Stay in the fight!",
              "push_intense": "[intense] Push through! Don't you dare coast!",
              "elapsed_serious": "[serious] Thirty seconds elapsed. Stay locked in.",
              "move_excited": "[excited] Let's GO! Move with a purpose!",
              "push_shouting": "[shouting] Push HARDER! You've got this!",
              "reset_commanding": "[commanding] Reset... breathe... attack again!",
          }

          os.makedirs("/tmp/domi-v3", exist_ok=True)

          for fname, text in PHRASES.items():
              payload = {
                  "text": text,
                  "model_id": "eleven_v3",
                  "voice_settings": SETTINGS,
              }
              data = json.dumps(payload).encode()
              req = urllib.request.Request(
                  f"{API}/v1/text-to-speech/{DOMI_ID}?output_format=mp3_44100_128",
                  data=data,
                  headers={"xi-api-key": KEY, "Content-Type": "application/json", "Accept": "audio/mpeg"},
              )
              path = f"/tmp/domi-v3/{fname}.mp3"
              try:
                  with urllib.request.urlopen(req, timeout=60) as resp:
                      audio = resp.read()
                      with open(path, "wb") as f:
                          f.write(audio)
                      print(f"{fname}.mp3 ({len(audio)} bytes) — {text}")
              except Exception as e:
                  print(f"FAILED {fname}: {e}")

          print("\nDone! 8 samples with Eleven v3 + audio tags.")
          PYEOF
      - uses: actions/upload-artifact@v7
        if: always()
        with:
          name: domi-v3-human-samples
          path: /tmp/domi-v3/
          retention-days: 1