Best of both worlds: Robust accented speech recognition with adversarial transfer learning

share