Product · n7kel Voice

n7kel Voice — speech-to-text and speech analytics in any language

The camera shows what an employee did; n7kel Voice shows what the employee said. The product transcribes conversations in the service area and analyzes them: was the script followed, were forbidden phrases used, was there a conflict, was an add-on offered. It works with any language; the price depends on the number of languages.

Interface illustration

Capabilities

What n7kel Voice does

  • 01

    Speech-to-text

    Conversation transcripts with speaker separation and timestamps. Languages are set per project — one or several.

  • 02

    Script compliance

    Greeting, clarifying the customer’s needs, stating the amount, offering a loyalty card, farewell — each dialogue shows which steps were completed.

  • 03

    Forbidden phrases

    Rudeness, unacceptable promises, phrases that contradict procedures. The client sets the list.

  • 04

    Conflicts

    Raised voices, an argument, a complaint — an event for the shift supervisor with a clip of the conversation.

  • 05

    Upselling

    Whether an additional product or service was offered and how the customer responded.

  • 06

    Conversation search

    Full-text search across transcripts: word, phrase, employee, site, period. Jump from a line to the clip of the recording.

Any language

n7kel Voice works with any language, including several in one conversation: for example, the employee speaks Russian and the customer answers in Kazakh or Armenian. Every utterance is tagged with its language, and the analysis rules — script, forbidden phrases, upselling — are set separately for each language.

The price depends on the number of languages in the project. For languages with local terms, product names and abbreviations, n7kel AI is trained further on the client’s examples.

Applications

Where it is used

How it works

From audio to report

  1. Audio source

    Microphones in the service area, cameras with built-in microphones, or call recordings provided by the client.

  2. On-site recognition

    Speech is converted to text on the client’s equipment. No external speech recognition services are used.

  3. Analysis

    The transcript is checked against the client’s rules: script, forbidden phrases, conflicts, upselling.

  4. Events and scores

    Violations reach the supervisor with a clip, and every employee and site gets a script compliance score.

  5. Link to video

    When used with n7kel Vakhta or n7kel Antifraud, a checkout event is shown together with what was said.

Scenarios

Typical scenarios

SituationWhat n7kel Voice does
The cashier did not state the amount or offer a receiptA flag for the missed script step and a dialogue score
Rudeness toward a customerA “forbidden phrase” event with a clip for the shift supervisor
Conflict at a service windowA “conflict” event; with n7kel Atlas connected, a review task
An add-on service was not offeredShare of dialogues with an offer, by employee and shift
Customer complaintSearch for all conversations by site and period, jump to the clip of the recording
A conversation in two languagesA language tag on every utterance, analysis rules for each language

Integration

Connection

  • existing microphones and cameras with audio, call recording systems, telephony;
  • transcripts and events passed to CRM, the service desk, BI and any client systems via the REST API and webhooks;
  • notifications to e-mail and Telegram;
  • joint operation with n7kel Vakhta, n7kel Antifraud, n7kel Fuel Station Control and n7kel Atlas.

Pilot

How we begin

  1. Languages and rules

    We agree on the languages, the script, the list of forbidden phrases and the evaluation criteria.

  2. Recordings

    Under a non-disclosure agreement, we receive sample conversations and show you the results on them.

  3. Pilot

    2–4 weeks at 1–2 sites, with employees and customers notified.

  4. Report

    Figures against the agreed criteria and recommendations on the script and staff training.

Questions

Which languages are supported?

Any. Languages are set per project; the price depends on how many there are. For local vocabulary and product names, n7kel AI is trained further on the client’s examples.

Does audio go to external services?

No. Speech is recognized on the client’s equipment, no external recognition services are used, and data is not shared with third parties, except where expressly provided for by the agreement and the law.

Do customers need to be notified?

Yes. Audio recording and analysis are used only when employees and customers are notified as the law of the country prescribes. We provide templates for signs and notices.

Hear how your customers are served

Tell us the languages and tasks, and we will show n7kel Voice on samples of your conversations. The price depends on the number of languages and is determined after the site survey.

The interfaces in the illustrations are schematic mockups with sample data.