1. Description

AMD determines whether a human is talking to you or a robot. It checks various voice parameters (Fig. 1) specific to the recording.

It consists of the Use AMD switch (Fig.1).

Fig. 1. AMD tab if Use AMD switch is off

The check is not performed if the switch is in the off position.

Recognition is done by checking various voice parameters specific to the recording. The parameters become available when the Use AMD switch is on.

If the Use AMD switch is on, the tab consists of the following elements (Fig.2):

1. Use AMD;
2. Artificial intelligence;
3. Maximum number of words;
4. Between words silence (ms);
5. Maximum word length (ms);
6. Minimum word length (ms);
7. Total analysis time (ms);
8. Silence threshold (ms);
9. After greeting silence (ms);
10. Greeting (ms);
11. Initial silence (ms);
12. Transfer NOTSURE to an agent.

Fig. 2. AMD tab if Use AMD switch is on

1.1.  Use AMD

The switch is responsible for enabling and disabling AMD autodetect.

- the autodetect is enabled;

- the autodetect is disabled (Fig.2 or Fig.3, corresponding to the position of the Artificial intelligence switch).

1.2. Artificial intelligence

The switch that is responsible for enabling and disabling artificial intelligence, according to the position of the switch:

- artificial intelligence is on;

- artificial intelligence is off.

If artificial intelligence is enabled, the tab consists of the following elements (Fig.3):

1. Use AMD;
2. Artificial intelligence;

3. Positive labels;

4. Playback file.

Fig. 3. AMD tab if  Artificial intelligence switch is on

1.2.1. Use AMD

The switch is responsible for enabling and disabling AMD autodetect.

 The description corresponds to the data in paragraph 1.1.

1.2.2. Artificial intelligence

The switch that is responsible for enabling and disabling artificial intelligence,

 The description corresponds to the data in paragraph 1.2.

1.2.3. Positive labels

Artificial intelligence can determine what we receive from the Subscriber into the following options:

• hyman;
• silence;
• ringback;
• voicemail - voice recording, e.g., answering machine.

You can choose which results you'd like to continue working within this field.

That is, when receiving the definition specified in this field from artificial intelligence, the connection switches to IVR.

The call ends when artificial intelligence has determined an option not selected in this field.

The History application displays how the call was recognized.

The determination takes place over time, the amount of which is configured separately on the side of artificial intelligence.

1.2.4. Playback file

You can select a file that artificial intelligence will play during recognition in this field. For example, a soundtrack can call the Subscriber to make a speech.

The file can be selected from the options available in the Media files section.

1.3. Maximum number of words

Here enter the maximum number of words in the greeting. It is considered a robot if more than indicated words are spoken.

1.4. Between words silence (ms)

The minimum pause (silence) between words is indicated to determine the following expression (in milliseconds).

1.5. Maximum word length (ms)

This field enters the maximum word length in milliseconds.

1.6. Minimum word length (ms)

Here enters the minimum duration of a continuous voice sound in milliseconds.

1.7. Total analysis time (ms)

Here enters the maximum time allowed to recognize if it is a human in milliseconds.

1.8. Silence threshold (ms)

Here enters the maximum duration of silence between words in milliseconds.

1.9. After greeting silence (ms)

Here enters the duration of silence after the greeting in milliseconds. If the entered value is exceeded, then consider a human.

1.10. Greeting (ms)

Here enters the maximum length of the greeting in milliseconds. If exceeded, then consider a robot.

1.11. Initial silence (ms)

Here enters the maximum duration of silence before the greeting in milliseconds. If exceeded, it is considered a robot.

1.12. Transfer NOTSURE to an agent

The switch is responsible for redirecting to an Agent of the Member that the auto detector could not recognize, according to the position of the switch:

- redirect to an Agent;

- do not redirect to the Agent.

When the switch is active, another switch is available - Transfer silence to an agent (Fig. 4).

The Transfer silence to an agent switch is responsible for redirecting the call to the Agent if the auto detector has made a definition of silence.

Fig. 4.  Transfer silence to an agent switch 

2. Opportunities

2.1. Change to autodetect options

2.2 Enable autodetect in the Queue

2.3 Enable redirection of unidentified Members to an Agent