emotion

false (default), true, xa

If set to true, Voci’s emotion detection feature uses a synthesis of acoustic features and word sentiment scores to determine if a given utterance is Positive, Improving, Neutral, Worsening, or Negative.

Overall emotion for a file may also be included in the top level JSON output. This value compares the emotion values in the first half of the transcript with the values in the second half.

Setting emotion=true also enables sentiment analysis.

The emotion=xa option uses increased accuracy emotion id (EID) to determine emotion rather than the older ASR engine. This option has a significant performance impact and should only be used when necessary. This option is only available using V‑Blaze version 7.1 or later.

Use rawemotion to include acoustic-only emotion.

Emotion detection is a licensed optional feature.

gender

false (default), true, xa

Set gender=true to activate gender detection. For each utterance, Voci will attempt to determine if the speaker is male or female.

The gender=xa option uses new gender ID (GID) technology rather than the standard emotion module. This option has a significant performance impact and should only be used when necessary. This option is only available with V‑Blaze version 7.1 or later.

Gender Identification is a licensed optional feature.

rawemotion

true, false, xa

When rawemotion=true, exposes all detected acoustic-only emotion information in JSON transcript output for each utterance as rawemotion.

The rawemotion=xa option uses new emotion ID (EID) technology rather than the standard emotion module. The new EID increases accuracy and expands classification to include HAPPY. This option has a significant performance impact and should only be used when necessary. This option is only available with V‑Blaze version 7.1 or later.

Possible return values are HAPPY, NEUTRAL, and ANGRY for V‑Blaze version 7.1 or later.

Possible return values are NONANGRY and ANGRY for versions prior to 7.1.

raw_events

true, false

Includes an additional raw_events list in the JSON output under utterances. This includes silence, filler words, wordex, and un-punctuated text. This parameter is useful for debugging and tuning and should not be used in the production flow.

This parameter does not process any substitutions.

sentiment

true, false

Sentiment is automatically enabled when emotion is enabled, but can be used independently of emotion.

When sentiment scoring is enabled, the ASR engine parses utterances for grammar; words that match certain grammatical models in those utterances are scored for sentiment. Detected sentiment is recorded in the JSON as sentimentex.

The emotion detection feature uses sentiment scores to determine if a given utterance is Positive, Improving, Neutral, Worsening, Negative, or Mixed (both Positive and Negative).

Unlike emotion, sentiment uses only linguistic information (words and phrases) to determine these values.

senti_list

filename , filename2...

The name(s) of the sentiment list(s) to be applied during transcription.

To use a custom sentiment rule with V‑Cloud, you must submit that rule to support@vocitec.com for integration.

Sentiment lists that can be specified using the senti_list tag must be placed in the /opt/voci/state/sentiment/ directory on each V‑Cloud server.

senti_rules

sentiment rule

Specifies a newline-escaped string that contains a set of manual sentiment rules. These rules do not require grammar parsing for the utterance to be scored. All words in an utterance are eligible for manual scoring with senti_rules.

Rules specified with the senti_rules tag are applied after any rules in the files identified by the senti_list tag. See senti_rules for additional information.