Eterna TeknolojiAkıllı Çağrı Yönetimi
TR

02Architecture

Six-layer architecture, three original modules

Audio leaves the cellular module digitally and is processed on the customer's edge computer. Nothing goes to the cloud.

fig. 02 · six-layer architecture
  1. L1

    Audio & Call Layer

    analog → digital

    SIM7600G cellular module answers the call; audio is passed to the edge computer digitally over PCM/I2S. Analog noise is eliminated at source.

    SIM7600G

  2. L2

    Speech Recognition Layer

    8 kHz → 16 kHz

    8 kHz phone band is adapted for the 16 kHz model; a quantised Whisper / wav2vec 2.0 model runs locally.

    Whisper-tiny

  3. L3

    Turkish Proper-Name Correction Layer

    biasing · phonetics

    Three stages: contextual biasing → Turkish-specific phonetic matching → dialog confirmation (confidence threshold 0.5).

    confidence ≥ 0.5

  4. L4

    Conversation & Workflow Engine

    YAML rules

    Required fields, question order and business rules live in YAML configuration; new sectors onboard without code changes.

    state machine

  5. L5

    Record Integrity Layer

    schema · enum · source

    Model output must conform to a predefined schema; enumerated fields are constrained; every persisted value must literally appear in the transcript; records cannot open with missing mandatory fields.

    pre-commit

  6. L6

    Speech Synthesis Layer

    Piper TR

    Piper Turkish voice model runs locally; XTTS-v2 is evaluated as a quality reference.

    on-device

output · structured record (YAML schema)

Original modules

These three components are not present in off-the-shelf models; they fill gaps left by the literature.

module · 01novel

Turkish-specific phonetic encoding

Phonetic coding algorithms exist for English (Soundex, Metaphone), German (Kölner Phonetik), Slavic (Daitmchotoff) and New Zealand English (Caverphone); no established counterpart for Turkish. The project fills that gap.

literature gap
module · 02novel

Free-form location → administrative unit resolver

The ambiguity created by identical neighbourhood names across provinces is resolved through in-dialog confirmation, without asking the caller administrative questions.

literature gap
module · 03novel

Pre-record integrity guard

Generative-model output is gated by schema constraints, enumerated value sets and source verification, preventing fabricated records.

literature gap