All functions are optional. Within a given function some components and procedures might be optional to still support the function but some will be required. Normative text in the following clauses thus describes requirements for support within an optional feature where it is desired to differentiate between optional and mandatory parts of the feature.

The MRFC shall request the MRFP to send tones to one, one of several, multiple or all parties connected in a call/session with a given tone identifier for each specific tone. The MRFC may request the tone to be played continuously until requested to be stopped. The MRFC may include in the request the length of time that the tone shall be played; the duration may be provisioned. The MRFC may then request a notification from the MRFP when the tone is completed. The MRFC may request DTMF detection while playing a tone. The MRFC may request that upon DTMF detection the MRFP stops playing a tone.

The function of playing announcement is to play audio media streams to the subscriber. The function can be used in services such as audio announcements, mail box services, play back recorded audio etc. The MRFC shall request the MRFP to play announcements to one, one of several, multiple or all parties connected in a call/session. The announcement may be referenced by identifiers that may be pre-configured, or dynamically obtained from the same MRFP for example due to Audio Record. The MRFC shall request sequences of predefined fixed announcements within one request to the MRFP. The MRFC may request announcements to be played in a loop until it commands the MRFP to stop. The MRFC may request the MRFP to play an announcement for a fixed number of times. The MRFC may request DTMF detection while playing an announcement. The MRFC may request the MRFP to stop playing an announcement when a DTMF digit is detected. The MRFC may request the MRFP to add the following variants to the announcements:

• Date/Day/Month
• Time
• Digits (the announcement may contain a number of digits to be controlled by the MRFC for example a telephone number)
• Money (currency)
• Integer (a value within the announcement that is controlled by the MRFC, e.g. "you are caller number 3 in the queue")
• Variants may have predefined default values for a given network.

The MRFC may request the MRFP to indicate when a specific announcement previously requested has been played successfully. The MRFP shall indicate error cases such as announcement not played successfully.

TTS (Text To Speech) is the process of automatic generation of speech output from text or annotated text input. The MRFC shall request the MRFP to play the text to one, one of several, multiple or all parties connected in a call/session. The text format shall comply with the SSML format as specified in [11]. The MRFC shall extract the SSML script from the VXML or other format XML script if received If the received text is another format than SSML, the MRFC shall generate a SSML script that may include the basic SSML text and the language type. The MRFC shall indicate to the MRFP the text-to-speech, by sending the SSML script or sending an URI reference to this SSML script. If the MRFC indicates the SSML script to the MRFP, the SSML text is sent inline in a H.248 command of Mp; the size shall be limited to avoid the segmentation in the Mp interface. The MRFC may remove unnecessary elements, such as the comments element, from the SSML document, providing that the result is a Conforming Speech Synthesis Markup Language Fragment as described in clause 2.2.1 of SSML ref [11]. This is however outside the scope of the current Mp specification work. If the SSML script size pre-processed results in segmentation in the Mp interface, the URI reference should be used. When the MRFC indicates the SSML script using an URI reference to the MRFP, two options can exist:

• the file (referenced by the URI) is located in the MRFP and it is a SSML text, hence the MRFP should play the text;
• the file (referenced by the URI) is located outside the MRFP; the MRFP may fetch the text and play it to the user otherwise the MRFP indicates an error.

The MRFP shall execute the basic SSML elements and may ignore the SSML elements not supported. The basic SSML elements include the root element "speak", language type and spoken text. The MRFC may request the MRFP to play a text in a loop until it commands the MRFP to stop. The MRFC may request the MRFP to play a text for a fixed number of times. The MRFC may request DTMF detection while playing a text. The MRFC may request the MRFP to stop playing a text when a DTMF digit is detected. The MRFC may request the MRFP to indicate when a text has been played successfully. The MRFP shall indicate error cases such as text not played successfully. Ignoring a non-supported SSML element shall not result in an error.

The function requirement of audio record is to record the audio media stream(s) and store it into a file. The function can be used in some services, such as the voice mail box service, conference service, etc. The MRFC shall request the MRFP to start the audio record from one or all parties connected in a call/session. If it is to record one party in a call/session, only the input stream of the party is recorded. If it is to record all parties in a call/session, the mixed stream of all parties is recorded. The MRFP file format shall comply with the 3GPP multimedia file formats as specified in the TS 26.244. The MRFC may request the MRFP to detect the DTMF digit while recording an audio. The MRFC may request the MRFP to stop recording and still retain the recording file. The MRFC may indicate to the MRFP the file format and the URI to store the recorded file or request the MRFP to return the record file URI. The MRFC may indicate to the MRFP the maximum record time. The MRFC shall request the MRFP to indicate the result and the cause of record completion when an audio has been recorded successfully. The MRFP shall indicate error cases such as audio not recorded successfully. The MRFC may indicate the MRFP to execute other functions, such as playing an announcement, when the MRFP is recording audio.

The MRFC shall request the MRFP to detect and report the DTMF digits. The MRFP shall report DTMF Digits detected as RTP Telephony Events (see RFC 2833) if the Telephony Event for DTMF Payload Type has been assigned to that interface. The MRFP shall report only single DTMF Digits.

ASR (Automatic Speech Recognition) function is that the recognizer processes the user input voice and may match that input against a target data to produce a recognition result that represents the detected input. In the IMS, the MRFP acts as the recognizer that is under control of the MRFC and finish the function of recognition. The MRFC shall request the MRFP to start the automatic speech recognition. The MRFC shall extract the SRGS recognition grammar script or URI from the VXML script if received or other format XML script if received. The grammar format shall comply with the SRGS format as specified in W3C Recommendation [12]. The MRFC shall indicate the SRGS script or the SRGS URI to the MRFP using H.248 packages. If the SRGS script is sent inline,. the size of the SRGS script shall be limited to avoid segmentation in the Mp interface. The MRFC may indicate to the MRFP the recognition mode: Normal Recognition Mode or Hotword Recognition Mode.

• If the MRFC indicates the Normal Recognition Mode to the MRFP, the MRFP shall attempt to match all of the speech against a recognition grammar and returns a no-match status if the input fails to match or the method times out.
• If the MRFC indicates the Hot-word Recognition Mode to the MRFP, the MRFP shall look for a match against specific speech grammar and ignores speech that does not match. The recognition completes only for a successful match of the recognition grammar or if the subscriber cancels the request or if the recognition time elapses.

The MRFP shall execute the recognition against the SRGS grammar and may ignore SRGS elements which are not supported. The MRFC may request DTMF detection while executing ASR. The MRFC may request the MRFP to stop ASR when a DTMF digit is detected. The MRFC may request the MRFP to indicate when a specific ASR has been completed successfully. When ASR is completed successfully, the MRFP may notify the MRFC the recognition result. The recognition result shall comply with a single recognition format (e.g. the EMMA format as specified in W3C Recommendation [13] or the NLSML format as specified in W3C Recommendation [15]). The MRFP shall indicate error cases such as ASR not executed successfully.