Feature List

Updated at:

The public cloud version of the outbound robot currently includes the following features:

Module /Component Name

Function point

Description

Manage outbound instances

Intelligent outbound service management

Manage all outbound instances. You can select the maximum parallelism of instances and add, delete, modify, and query instances.

Task management

Create a task

If you want to create an outbound call task, you can use a process template, configure the robot voice, select a task triggering mode (manual or automatic), and specify phone numbers, number of channels, and number of retries.

List Management

You can upload an outbound call list by using a template or manually upload a custom list.

Start /Pause /Stop Task

You can start a pending task or a suspended task, or suspend an ongoing task.

Add a stop-call list

You can add a list of customers that are not called.

Scheduled task execution

You can configure a scheduled task by setting the start time and the time period.

Task priority and concurrent scheduling

Supports urgent and daily priority settings for tasks, and intelligent scheduling based on the specified priority. You can also set the concurrency of a task. When the task is running, the concurrency configuration value is preferentially met. If an instance has idle concurrency, it is randomly assigned to a running task to improve task dialing efficiency.

Ringing Duration Control

You can configure the ring duration for outbound calls. You can hang up calls that are not received and have reached the configured duration. This improves the efficiency of outbound calls. And combined with historical data to provide a default optimal value.

Redial Policy

You can configure the redial policy for missed calls, including the gap, number of redials, and whether to call out in case of empty number downtime and overdue payment.

Local Number Priority Policy

Support city number priority, province number priority dialing policy selection, intelligent matching according to the call and call home, improve the number connection rate level. The number library can be managed in the product background.

Real-time monitoring of outbound tasks

Shows the progress of tasks in real time, including task execution statistics, execution details, and call records. For example, there are 150 total tasks, 83 outgoing calls, and the distribution of ringing tone duration.

SMS Sending

Supports sending text messages.

Export the result of an outbound task

You can download all completed call results, including statistical results, outbound call logs, recording files, and call records. You can export only the text and statistics of recording files, and push the result information.

Copy Tasks

You can quickly copy existing outbound tasks and inherit the number.

Outcall task details

Task details list

Provides an overview page for outbound call task details, including the progress of all outbound call tasks under the task, the execution of the last call, and more details, such as the call duration, the end of the call, the reason for not connecting, the specific reason for call exceptions, and the variable information collected from the call.

Filter and search

The list supports the customer's phone number, call status, whether there is an answer, Whether to reject the on-hook, whether to complete and other conditions to search.

Statistical Report

Includes task execution statistics, top 30 intent and node ranking, funnel distribution, and variable distribution. The call flow optimization data presentation mode supports full filling and presentation of all columns in the call flow form, making it easier for businesses to perform flexible data analysis based on their own demands.

Call content query

You can listen to the recording and view the escape results. You can analyze and obtain the customer's intention, conversation rounds, call duration, reason for hanging up, the party that hung up, and other basic information about the call in real time.

Support end-play display when hanging up

The session history shows whether the playback finished before the call hung up.

Call recording download

Support recording timely save, play online and download playback.

Ring tone download

Support ring tone playback and download (background download).

Scene management

Management Scenarios

Manages the scenario of the current outbound instance and associates it with the corresponding robot scenario.

Pre-synthesis mode

Pre-composes the speech in the robot conversation into a recording using TTS to reduce the time period for calling in outbound calls.

TTS Settings

You can use TTS to play the video. You can select the tone color, and configure the speed, intonation, and volume. You can listen to the demo and download the demo.

Apsara Stack Resilience settings

You can configure Apsara Stack Resilience acoustic models, language models, and hotwords. In scenario management, you can select and load models and hotwords in the self-learning platform to help you perform personalized speech recognition and tuning for different industries.

Intelligent Undertaking Language Management

Switch, configure, and intervene in the strategy for each language.

Graceful break switch

Turn the graceful interruption capability on or off for this scene

emotion recognition

Supports emotion recognition capabilities that integrate speech and semantic features to more accurately determine a variety of emotions on the user side.

Intelligent broken sentence switch

Enable or disable intelligent sentence breaking (large gasp detection) for the scene

Softphone (voice test)

Supports end-to-end voice testing by using soft phones on the product side to simulate actual call scenarios in all aspects, and supports special requirements such as variable control, voice detection, tone acceptance, key input, and silent interruption during calls. It provides more intuitive and authentic testing channels for customers, facilitates troubleshooting and tuning of process and speech problems, and ensures end user call experience.

Outcall History

Outcall History

Query all historical sessions based on Elasticsearch. You can search for multiple fields such as called number, calling number, time, task, and scenario. Supports downloading export details.

Number management

Management Number

Manage the number of the current outbound instance. Allows you to configure the number, throttling period (seconds), and throttling times.

Data labeling

Labeling Tasks

You can extract call data to create labeling tasks based on scenarios, call results, and call time ranges. Supports proportional and quantity sampling.

Task labeling

Supports semantic recognition results, similar intent questions, Apsara Stack Resilience translation, hot word lists, training sets, and call data tagging. It also supports full-screen fast labeling.

Task Statistics

Supports statistical analysis of speech and semantic annotation results and call label statistics.

Download Labeling Results

Support call result data download.

Dataset management

Apsara Stack Resilience a popular hotword vocabulary

In the speech recognition service, if the default recognition effect of some words in your business field is not good enough, consider using the hotword feature to add these words to the vocabulary to improve the recognition result. Such as place names, names, specific brand names, etc. These hotwords can be loaded into scenes and tasks to improve recognition.

system administration (anti-harassment strategy)

Global dial window restrictions

The time limit for outbound calls for all tasks under the outbound service, including the custom time window and whether only on weekdays

Called Number Daily Maximum

Limit the maximum number of outbound calls per day for the same called number.

Blacklist management

Allows you to configure outbound blacklists, including phone numbers and comments.

Whitelist management

Allows you to configure an outbound whitelist, including phone numbers and remarks.

system administration (Other)

Concurrency configuration

The number of concurrent lines for outbound calls is automatically configured.

Voice self-learning platform configuration

You can configure and verify the cluster addresses of multiple voice algorithm services, including all Voice Messaging Service such as Apsara Stack Resilience, TTS, and self-learning platform.

Delay playback for N seconds after switching on

Allows you to configure the N-second delay playback policy.

VUI logic

Interrupted

You can configure a policy that allows robot speeches to be interrupted by customer speeches. When the robot is interrupted by the customer, the robot stops playing the messages and starts to process the user speech.

Silent reply and configuration

You can configure a policy to trigger a silence timeout speech and set the actions upon consecutive silence periods.

Reject Reply

You can configure a rejection policy to trigger a speech and set the speeches and actions upon consecutive rejections.

Silence detection configuration

Allows configuration.

Key input

You can use the speech and the key dual tone multi-frequency (DTMF) key input modes.

VAD adjustment

You can customize the threshold for noise filtering in a scenario. You can adjust the threshold based on the actual scenario to improve the call effect.

speech semantic integration ability

Phonetic semantic integration-

Intelligent acceptance language

An intelligent receiving language capability, which can judge according to the content type described by the user and automatically add colloquial receiving words, such as "uh-huh" and "ok", before DS reply content. This ability to undertake language will bring the following advantages:

-The end-to-end latency experience in the phone is expected to be reduced by about 600ms, which will be an important differentiator.

-More natural conversation experience

-The model pre-packages 10 kinds of receiving languages, and the pronunciation and intonation are fine-tuned and ready to use out of the box.

-Targeted optimization of the connection between the acceptance language and the subsequent TTS synthesis content to ensure natural and smooth connection

Phonetic semantic integration-

Graceful interruption

The new algorithm is used to greatly optimize the false interruptions caused by noise, background voice, acceptance language, and non-interruption intention. No manual intervention is required and one-click switch is required.

Phonetic semantic integration-

Intelligent sentence breaking (large gasp detection)

The new model is used to judge the sentence segmentation, and the acoustic and semantic features in the user's expression process are combined to judge whether the user has finished expressing the sentence. If the user pauses for too long, the robot will automatically generate the following words such as "Well, you continue" and "Well, then". It is a comprehensive upgrade to the traditional use of VAD silent duration logic.

Mobile phone number interactive collection

Support the collection and confirmation of mobile phone numbers in real person mode and through batch communication for many times, support for reading numbers in any segmented mode, support for modification, re-collection, secondary confirmation, etc.

Other functions

Batch testing tools

Batch online testing of the new dialogue design, output test results and modification suggestions.

Session records

Session record storage: Stores session audio and other session log entry. Audio recordings are retained for at least 6 months.

Endpoint Detection

The process of analyzing the input audio stream to determine the start and end of the customer's speech.

Noise elimination

With efficient noise cancellation capabilities, reduce the noise of customers in the actual environment, the maximum retention of the user's original voice information.

Support button number collection

Support button receiving function.

Processing of ultra-long reply content

If the node reply is divided by carriage return, it can be automatically divided into multiple TTS segments, thus supporting the broadcast of ultra-long reply content.

Segmented recording

Segmented recording is supported in the system administration. After you enable this feature, you can play the user voice in call recording in the terminal. This way, you can quickly obtain the voice information of the node.

Visual dialog flow editing

Provides a variety of standard processes in classic scenarios. You can add, delete, and edit process nodes, modify node jump relationships, and drag and drop process node layouts through visual editing of policy processes.

System background

SIP Line Registration Management

Supports quick connection to SIP lines, manages the status of SIP lines and bound numbers, and supports multi-tenant isolation.

Transfer to manual policy management

You can configure and encapsulate manual transfer policies and call the corresponding policies in dialog design.

double center

Dual-center hot standby

Supports the dual-hub hot standby capability, which is available only in the Advanced Edition.

Secure encryption

Clear Text Password Encryption

The outbound backend encrypts data such as passwords to meet customer security compliance requirements.

Conversation content slot encryption

Supports encryption settings for slots, implements database drop encryption on the outbound platform, and supports coding when displaying and exporting conversation logs.