Whole call recording (Revisited)

About four years ago I posted an inquiry on the (now expired) Cisco Dev Forum asking about Cisco VXML platform's support for "whole call recording" (WCR), and the answer I got then from the gateway team was "No, WCR is NOT supported by the Cisco VXML gateway." See the following link for an archived copy of that conversation:

http://solutionpartner.cisco.com/web/vgapi/community/-/message_boards/message/2015364?_19_delta=20&_19_keywords=&_19_adv…

Four years later, I'm hoping to revisit the same discussion to see anything changed in the gateway's support for WCR.

Specifically, here are my questions:

1) Is there a way to turn on Whole Call Recording in the VXML platform itself? (Please note that WCR is a feature supported by some VXML platforms out there including Voxeo and Tellme).

2) If the answer to 1) is no, then is there a configurable way to force the IVR to send all prompt requests (both pre-recorded audios and TTS) via MRCP to Nuance, so that Nuance would play ALL the prompt audios, AND be able to capture all of the system prompts in the Nuance Whole Call Recording files?

Please note that based on my wireshart test results so far, it seems that by default ONLY TTS prompts are being passed (from the IVR gateway to Nuance), but NOT the pre-recorded audios (so Nuance WCR files can only capture TTS prompts, but NOT pre-recorded audios).

Your timely response to questions 1) and 2) would be appreciated!

Thanks,
Rachel

Hi Rachel,

Cisco VoiceXML supports HTTP, TFTP, RAM, RTSP, and flash memory as sources of prerecorded audio prompts for playout. The audio files can be played out in cached or buffered mode, or they can be streamed.

For VoiceXML Recording please refer below link for more information.

http://www.cisco.com/c/en/us/td/docs/ios/voice/vxml/developer/guide/vxmlprg/refgde3.html#wp1035789

Thanks,

Raghavendra

Hi Raghavendra,

Thanks for the quick response!

My question was not about the gateway's support for the standard VXML <record> functionality, but rather I was asking about a Whole Call Recording capability that would record the two-way call interactions between the caller and the IVR systems from beginning to end of the IVR call session, including ALL of the following:

• Prompts played (both Pre-recorded and TTS)
• DTMF inputs
• Speech inputs and transcription
• turn-taking and latencies in between

Here again are my questions:

1) Is there a way to turn on Whole Call Recording in the Cisco VXML platform, without using any 3rd-party recording tool?

2) If the answer to 1) is no, is there a configurable way to force the IVR to send all prompt requests (both pre-recorded audios and TTS) via MRCP to Nuance (so that Nuance would be able to capture all of the system prompts in the Nuance-generated Whole Call Recording files)?

My wireshark tests against Nuance Speech Server suggests that by default the gateway ONLY passes TTS prompts to Nuance, but NOT pre-recorded prompt audios.

Thanks,

Rachel

Hi Rachel,

i am checking with Dev Team on this, i will get back to you once i got answers for your questions.

Thank,

Raghavendra

Hi Rachel,

please find the below response from the Dev Team.

1)Is there a way to turn on Whole Call Recording in the VXML platform itself? (Please note that WCR is a feature supported by some VXML platforms out there including Voxeo and Tellme).

- We don’t have support for this.

2)If the answer to 1) is no, then is there a configurable way to force the IVR to send all prompt requests (both pre-recorded audios and TTS) via MRCP to Nuance, so that Nuance would play ALL the prompt audios, AND be able to capture all of the system prompts in the Nuance Whole Call Recording files?

- There seems to be some confusion with the terminology you are using here. I think you want to ask if there is some way to send the audio to NUANCE server such that NUANCE server will record the audio. Is that right?

Using MRCP you can do the recording. However, this is used only for application specific recording. I have not seen any use it for "call-recording".

Thanks,

Raghavendra

Hi Raghavendra,

Please see my follow-up questions/comments inline:

1) Is there a way to turn on Whole Call Recording in the VXML platform itself? (Please note that WCR is a feature supported by some VXML platforms out there including Voxeo and Tellme).

[Cisco]: - We don’t have support for this.

[Rachel]: As WCRs offer the most effective way to provide a REALISTIC and DETAILED picture of the actual caller experience,

particularly when it comes to turn taking, latency, barge-in, background noise and prosodic features of caller input, it's no surprise that some browsers such as Voxeo have decided from years ago to provide WCR as a builtin capability (as in:

http://docs.voxeo.com/tropo/2.0/frame.jsp?page=recording_entire_call.htm). So, my follow-up question is: Any plans in your near-term roadmap to add support for this in the VXML browser?

2) If the answer to 1) is no, then is there a configurable way to force the IVR to send all prompt requests (both pre-recorded audios and TTS) via MRCP to Nuance, so that Nuance would play ALL the prompt audios, AND be able to capture all of the system prompts in the Nuance Whole Call Recording files?

[Cisco]: There seems to be some confusion with the terminology you are using here. I think you want to ask if there is some way to send the audio to NUANCE server such that NUANCE server will record the audio. Is that right? Using MRCP you can do the recording. However, this is used only for application specific recording. I have not seen any use it for "call-recording".

[Rachel]: What I'm trying to ask in my 2nd question is whether or not the Cisco VXML browser provides a CONFIGURABLE option, presumably in the IVR application, to force the gateway send ALL prompt contents (including both TTS and pre-recorded audios) to Nuance (for that particular IVR). As we know, the browser anyways has to send TTS prompts (e.g. for dynamic contents) to Nuance TTS engines for TTS prompt playing, which means a TTS license will HAVE have to be tied up for the duration of the entire call session. So if the browser also sends pre-recorded prompt audios for Nuance TTS to process, it is not going to tie up any additional license. The only potential impact i can think of would be slight latency or additional stress on Nuance TTS. Either way, i was just wondering whether or not that option (of letting Nuance process both TTS and recorded audio prompts) exists (at all) with the CVP browser. Let me know.

Thanks,

Rachel

Hi Rachel,

I don't think any plans as of now, if there is any will let you know.

Thanks,

Raghavendra