If a presentation in Narrator uses characters with voices supplied both by VoiceOver (bundled with Mac OS X) and by Cepstral, it is possible that the parts using VoiceOver voices could return unintelligible sounds upon playback within Narrator.

With regards to this, we have noted the following:

• Using VoiceOver voices exclusively for character voices does not reproduce this behavior.
• Using Cepstral voices exclusively for character voices does not reproduce this behavior.
• Exporting to .AAC or directly to iTunes does not reproduce this behavior.

This is a known issue. We intend to resolve it in a future release of Narrator.

As a workaround, until we have resolved the issue, we suggest using either VoiceOver Cepstral voices exclusively (without mixing the two) for presentations that will not be exported to .AAC or iTunes.