His face looked scary, but his voice was gentle and calm.

Literal

His face [topic-は] scary-seeming appear opposite, voice [topic-は] gentle calm was.

怖そうに見える ('looks scary') uses ~そう (appearance) + に見える ('appears'). 反面 (はんめん, 'on the other hand, conversely') is the pivot word connecting the contrast. おだやか (穏やか, 'calm, gentle') describes the voice. The sentence uses double は — one for face, one for voice — cleanly contrasting two aspects of the same person.