Cisco Unified MeetingPlace, Release 6.x -- Guidelines for Recording Prompts
Use the following guidelines when you record prompts and create prompt sets for Cisco Unified MeetingPlace 8100 series.
- Recording environment. Make sure the environment is extremely quiet. Do some sample recording beforehand and listen to the result. The recorded passages should be free of hissing, pops, frequency distortion, and other types of noise.
- Silence before and after each prompt. Each prompt should have between 50 to 100 milliseconds of silence at the beginning and end. Too little silence creates pops when the prompt is concatenated with other prompts (played just after or before other prompts). Too much silence creates awkward gaps in overall speech when several prompts are played together. In some cases, 100 milliseconds might not be enough to eliminate pops at the beginning and end of concatenated prompts. In this case, try increasing the silence to 150 milliseconds. To measure the amount of silence at the beginning or end of a prompt (or to see if there is silence), bring up the prompt in Adobe Audition or a similar voice editing program. Open the file as a 16-bit, 8-kHz .wav or raw PCM file. Adobe Audition allows accurate measurement of silences passages.
- Format. Record all prompts in linear PCM format, which is 8-kHz sample rate with 16-bit samples. Prompts should be in headerless format. If not, then each must be processed in Adobe Audition or CoolEdit to strip the headers.
- Level. The nominal recording level for non-silence sections should be -24 dBm. Avoid very loud recording levels because they can cause clipping, which creates clicks and pops in the middle of the prompts.
Note: If you upgrade your Cisco Unified MeetingPlace Audio Server to a newer release, you may need to reinstall all your custom prompts. Back up your custom prompts using the bnrprompt -B command.