An agent uploads a greeting via the Talk UI. It sounds crisp over the UI, but when someone calls in and hears it, it sounds grainy.
The reason why the greeting sounds less clear is because it has been compressed. Twilio has some suggestions for creating recordings so they do not lose quality when they are played over a line. Here are their Best Practices for Audio Recordings.
Key tips from that article:
Transcode your audio file in advance. If your audio software editor has sample rate converter and encoding capabilities, this option affords you some degree of control over the final results. Also, the quality of sample rate converters (to go from high-quality 44.1Khz or 48Khz to 8Khz) varies depending on the algorithm used, so you can compare Twilio with that of your own audio software.
Words of wisdom from Wesley on the Talk Dev team: Use audacity to encode/sample at 8Khz so it's close to what will get pumped through the phone lines. That way you won't get unexpected clipping and other nasty surprises when Twilio transcodes it for you.
Since Twilio encode in G711 M ULAW, and from testing I can confirm that the recommended quality is mono (as stereo unnecessarily doubles the file size) and encoding as follows: