Connector settings
The Connectors settings section is a form-based interface to guide you as you specify connector credentials and processing properties for files. The settings vary depending on the connector type.
engine, vertical model, locales, diarization, redactions, substitutions and other values can be defined using Connector settings, currently, there's no merge mechanism between the Speech API payload and connector settings, so Experience Cloud will use the the values defined in the Speech API payload if present. - Credentials
- Authentication credentials to access the external system where the files are stored. See Credentials. Note: The remaining tabs in the connector editor cannot be selected until you select or create a valid set of credentials.
- Connection ID
- Unique identifier of the connection profile. This setting is only required for push connectors (Red Box).
Red Box pushes a JSON file to an MMFT along with an audio file. With each implementation, Red Box will provide the connection ID so that the right instance picks up the metadata file. Unlike with other types of connector, metadata and audio files can arrive in any order.
With each implementation, LivePerson will provide the connection ID so that the right instance picks up the metadata file.
For more information see Get started with Medallia Speech.Note: Push connectors, such as Red Box connectors, don't use credentials but rely on a connection identifier for their connection profile. For information about this parameter see Get started with Medallia Speech and Connector settings. - Vertical Model
- Medallia Speech vertical model to use for analyzing the contents.
- NICE Call Recorder
- Recorder type defined in NICE UserHub for the calls you want to retrieve. NICE supports two different types of Call Recorders: CXOne and ACD.
Call recorders are storage areas where audio data can reside.
- Directory
- Directory on the S3 bucket or the SFTP to download and import audio files.
This setting is only required for file-based connectors (S3 buckets, SFTP, and Five9).
- Metadata directory
- Directory on the S3 bucket or the SFTP to download and import metadata files.
This setting is only required for file-based connectors.
Restriction: Beginning with the 2026 H2 release, Experience Cloud only supports separate directories dedicated to each file type for signal and metadata files. Existing connectors are not restricted to separate directories and it is technically feasible to use one singular directory for both file types, but this can make testing difficult. For existing connectors with a single directory, apply the metadata file filter to assist the connector in identifying metadata files. Make metadata file names obvious as to their purpose to enable easy setup of regex filters. - Metadata file filter
- Filter to match metadata files in the metadata directory, for example: *.csv.
This setting is only required for file-based connectors.
- Integration Type
- Specifies whether the source data includes original audio and metadata, redacted audio and metadata, or redacted audio, redacted transcripts and metadata.
- Engine
- The speech-to-text transcription engine to use for the call. Available options are:
-
Voci Engine
-
Amazon Transcribe Engine
-
- Language Identification
-
Restriction: This feature is inactive by default. Ask your Medallia expert to file a Feature activation request with Medallia Support.
This option is available with the Amazon Transcribe engine only. Enable this option to identify the language in speech data prior to transcription. When Language Identification is enabled, use the Language Options dropdown to select the languages the engine can detect. Select All to detect all available languages. Multiple languages may be selected.
The languages available and displayed in the dropdown are limited to those covered by the Amazon Transcribe engine and supported by Medallia Text Analytics. Languages from the Medallia Voci engine are not supported.
- Custom Vocabulary
-
Custom vocabulary (also known as out-of-vocabulary or OOV) is an ASR tuning feature for the Voci engine that improves transcription accuracy for audio that contains brand- and industry-specific terminology. Custom vocabulary supports all Voci language models and enhances those language models with new words and preferential treatment for those words.
To learn more, see Custom vocabulary.
- Transcription Substitutions
- Set of transcription substitutions to make during processing.
Substitutions can correct errors in transcripts using substitution rules that find and replace transcription errors with corrected values. For instance: replace "appeal box" with "a PO Box" or "triple A" with "AAA".
The format is that of a JSON object, where the keys are the original versions to find and the values are the replacement versions. See the example below for proper formatting:
{"appeal box":"a PO box","triple A batteries":"AAA batteries"} - Customer Locale
- Primary language used by the customer during the exchange.Note: When using the Voci engine, if the field Agent Locale has a value, then customer locale will be used for customer channel and the agent locale for the agent channel. Otherwise, Customer Locale will be used as media language for the entire file.
- Agent Locale
- Primary language used by the agent during the exchange.Note: This parameter is only available for the Voci engine. The rest of the engines only support the parameter Locale, which is the primary language for the exchange.
- Apply Diarization
- Determines whether diarization needs to be applied to the audio file during processing. Diarization presumes two people are speaking, and separates mono audio recordings into distinct channels by categorizing speech into two groups. So, this setting only applies to mono-channel recordings, which need to be diarized.Restriction: Diarization is not available for the Amazon Transcribe engine.
- Apply Redaction
- Determines whether redaction is performed on the exchange. It specifies if numeric personally identifiable information such as social security numbers and phone numbers, should be redacted from the file.Note: For security purposes Experience Cloud automatically redacts credit card numbers, Social Security Numbers, and street addresses from the transcription and playback audio. If your company wishes to keep that information visible, set this field to "No".Restriction: Redaction is not available for the Amazon Transcribe engine.
- Redaction configuration
- This option is available with the Voci engine only. Set of transcription rules to be applied for Medallia Speech processing. For information, see Speech redaction syntax and default rules. Tip: This configuration parameter is optional: standard redaction configuration applies if you do not enter any rule.
- Music detection
- This option is available with the Voci engine only. When you enable music detection, all utterances will be passed through an algorithm to be classified as music or non-music. Utterances classified as non-music will be handled as normal. Utterances classified as music are assumed to contain noisy audio and will not be transcribed.Restriction: This feature is inactive by default. Ask your Medallia expert to file a Feature activation request with Medallia Support.Important: Utterances classified as music will not be processed by any optional transcription features such as LID, GID, EID, and diarization. Optional transcription features ignore any utterance classified as music.
- Enable emotion processing
- This option is available with the Voci engine only. Enable this option to apply machine learning techniques to the analysis of emotion, which is determined from the acoustic features of speech. For information, see Emotion analysis.
- Agent Channel
- Determines which of the 2 channels (0 or 1) is associated with the agent. The other channel is associated with the customer.
The initiator of a call is assigned to channel 0. For inbound calls, set the agent channel to 1. For outbound calls, set the agent channel to 0. Confirm how your telephony system records data to audio channels to properly set this value.
This setting is only required for file-based connectors.
- Split Segments
- Determines whether the call should be split into segments. This approach is used to manage calls with multiple units, for information see Multi-touch programs.
This setting is only required for Genesys PureCloud connectors.
For a singular customer interaction, Genesys stores one audio file and one conversation ID, even if the customer is transferred among multiple agents (units). A segment represents an interaction between a customer and an agent. Experience Cloud represents each call segment as an individual record.
Call ID Segment ID Experience Cloud Call ID Start time End time Agent ID 1234 — 1234 0:00 15:06 Agent 2 1234 1 1234_1 0:00 4:36 IVR 1234 2 1234_2 4:36 10:11 Agent 1 1234 3 1234_3 10:11 15:06 Agent 2 -
There is a master record that captures the full call encompassing all segments.
-
Each segment is also created as a record in Experience Cloud.
-
The last agent that interacted with the customer is the unit assigned to the full call record.
-
The Genesys API provides full call audio recordings, but the connector uses the API information to split the call based on segment start and end times.
-
- Alert failure threshold
- Failure rate (expressed in percentage values) that triggers an event ticket. For information see MEC_CONNECTOR.
For example: if a connector run attempts to import 100 records and the alert failure threshold is 5%, when the connector run fails for 5 or more records on that job, an event ticket is triggered.
