register-twilio-test-audio

Use when adding new test audio files for Twilio voice calls, uploading audio to S3, or updating the twilio_place_call.py script with new audio options.

$ インストール

git clone https://github.com/cncorp/arsenal /tmp/arsenal && cp -r /tmp/arsenal/dot-claude/skills/register-twilio-test-audio ~/.claude/skills/arsenal

// tip: Run this command in your terminal to install the skill


name: register-twilio-test-audio description: Use when adding new test audio files for Twilio voice calls, uploading audio to S3, or updating the twilio_place_call.py script with new audio options.

Register Twilio Test Audio

Upload audio files to S3 and register them for use with twilio_place_call.py.

Prerequisites

  • SSH access to server-local with AWS CLI configured
  • Audio file must be in mulaw format (8kHz, mono) for Twilio compatibility

Quick Reference

ItemValue
S3 Bucketcodel-development-tts-audio
S3 Prefixtest-audio/
URL Patternhttps://codel-development-tts-audio.s3.amazonaws.com/test-audio/{filename}
Scriptapi/src/scripts/twilio_place_call.py
DictAUDIO_FILES

Steps

1. Convert audio to mulaw format (if needed)

ffmpeg -i input.wav -ar 8000 -ac 1 -acodec pcm_mulaw output_mulaw.wav

2. Upload to S3 and make public

From server-local (uses --profile codel):

# Upload single file
aws s3 cp /path/to/audio_mulaw.wav s3://codel-development-tts-audio/test-audio/ --profile codel

# Make it publicly readable (required for Twilio to access)
aws s3api put-object-acl --bucket codel-development-tts-audio --key test-audio/audio_mulaw.wav --acl public-read --profile codel

# Verify upload and public access
curl -sI "https://codel-development-tts-audio.s3.amazonaws.com/test-audio/audio_mulaw.wav" | head -1
# Should return: HTTP/1.1 200 OK

3. Update twilio_place_call.py

Add entry to AUDIO_FILES dict:

AUDIO_FILES = {
    # ... existing entries ...
    "newname": "https://codel-development-tts-audio.s3.amazonaws.com/test-audio/audio_mulaw.wav",
}

4. Test the audio

# Verify URL is accessible
curl -sI "https://codel-development-tts-audio.s3.amazonaws.com/test-audio/audio_mulaw.wav" | head -1
# Should return: HTTP/1.1 200 OK

# Test with a call
docker compose exec api python src/scripts/twilio_place_call.py --audio newname --to '+1234567890' --duration-minutes 1

Common Issues

IssueSolution
403 ForbiddenCheck S3 bucket permissions, file may not be public
Audio doesn't playEnsure mulaw format (8kHz mono)
Call gets errorVerify URL returns 200, not 503

Naming Convention

Use descriptive names with _mulaw suffix:

  • fight_mulaw.wav - Conflict/argument audio
  • neutral_mulaw.wav - Normal conversation
  • healthyfight_mulaw.wav - Healthy disagreement