Productivity glossary
Glossary of productivity, time management, & work-life balance terms.
A
AI Voice Agent
An autonomous conversational system capable of conducting human-like, two-way spoken phone calls over PSTN or WebRTC, understanding intent, answering questions, and executing tasks like scheduling or lead qualification.
ASR (Automatic Speech Recognition)
The speech-to-text subsystem that converts incoming telephone audio into text tokens in real time, accounting for acoustic noise, accents, and dialect variations.
Answer-Only Billing
A billing model where platform minutes are metered strictly from the moment the callee answers the line; ring duration, voicemail drops, and failed connection attempts incur zero cost.
Acoustic Endpointing
The real-time audio analysis used to determine when a human caller has finished an utterance, preventing the AI agent from prematurely cutting in during natural pauses.
B
Barge-In (Interruption Handling)
The capability of an AI calling agent to immediately stop speaking and listen when a human caller interrupts mid-sentence, eliminating robotic monologue behavior.
Backchannel Detection
Acoustic filtering that recognizes passive conversational feedback (such as 'uh-huh', 'okay', or 'hmm') without treating it as an interruption, allowing the agent to continue speaking naturally.
Bidirectional CRM Write-Back
An automated data synchronization loop where contact records are ingested before dialing and updated post-call with recordings, transcripts, intent scores, and booking details.
Budget Pacing
An outbound dialing safeguard that throttles calls-per-minute and concurrent channels to ensure campaigns stay within pre-allocated daily spend and minute ceilings.
C
Call Automation
The end-to-end orchestration of outbound dialing campaigns, inbound call routing, qualification logic, and CRM updates without manual human agent intervention.
Code-Mixing (Hinglish)
The linguistic practice of alternating between two or more languages - predominantly Hindi and English - within a single utterance or conversation, standard in Indian telephony.
Concurrency (Concurrent Calls)
The maximum number of simultaneous live telephone channels an account or campaign can maintain at any single second.
Conversational AI
A set of technologies including LLMs, natural language understanding, and voice synthesis that enables computers to engage in fluid, contextual, multi-turn dialogue.
D
Dead-Air Watchdog
A background monitor that checks whether a caller remains silent after answering or mid-call, triggering a conversational re-engagement prompt ('Are you still there?') before disconnecting.
DNC (Do Not Call) Suppression
The automated exclusion of phone numbers listed on internal opt-out lists or national regulatory registries prior to campaign queueing and dispatch.
Double-Talk Detection
A signal processing technique that detects when both the caller and the voice agent are speaking simultaneously, preventing echo feedback loops and prioritizing human speech.
DTMF (Dual-Tone Multi-Frequency)
The telecommunication signaling system that translates keypad button presses into audible tones for IVR menu navigation and numeric data input.
E
Echo Cancellation
A signal processing algorithm that removes the agent's own transmitted voice from the incoming telephone microphone feed, preventing the agent from interrupting itself.
Endpointing (VAD)
Voice Activity Detection algorithms that measure spectral energy and silence duration to accurately segment human speech into discrete conversational turns.
F
First-Utterance Latency
The measured time duration between the telephone network signaling 'answered' and the AI voice agent delivering its opening greeting to the callee.
Full-Duplex Audio
A bi-directional communications channel that allows simultaneous two-way transmission and reception of audio, essential for natural interruptions.
H
Hinglish Adaptation
Specialized language modeling calibrated to comprehend vocabulary, syntax, and colloquialisms blending Hindi and Indian English seamlessly.
HubSpot OAuth Connector
A native integration module that authenticates directly with HubSpot to read lead lists and write back call outcomes, notes, and call recordings.
I
Inbound AI Receptionist
An always-on voice agent assigned to an incoming phone number that answers calls immediately, triages customer requests, answers FAQs, and routes callers.
Intent Classification
The machine learning process of identifying the underlying goal of a caller's statement (e.g., booking a demo, requesting a refund, asking for pricing) from natural speech.
K
Knowledge Base Ingestion
The process of indexing company documents (PDFs, text files, product sheets) so an AI calling agent can ground factual answers in verified business data during live calls.
L
Lead Qualification
The systematic extraction of prospect criteria - such as budget, authority, need, and timeline (BANT) - during a phone conversation to determine sales readiness.
LLM (Large Language Model)
The foundational reasoning and generative engine that processes transcribed speech tokens, evaluates context, and drafts conversational responses.
O
Outbound Dialler
An automated system that dispatches telephone calls to prospect lists according to predefined pacing rules, scheduled time windows, and concurrency thresholds.
P
PSTN (Public Switched Telephone Network)
The global aggregate of circuit-switched telephone networks that connects traditional mobile cellular lines and landlines to voice servers.
Persona (Voice Persona)
A synthesized vocal identity combining specific acoustic timbre, speaking rate, tone, and regional accent characteristics (e.g., Aarav, Priya, Rohan).
Pre-Call Worker Warmup
An optimization where model sessions and audio pipelines are initialized while the phone is ringing, ensuring zero lag when the callee answers.
S
SIP Trunking (Session Initiation Protocol)
The voice-over-IP signaling protocol that links CallUltra's cloud infrastructure to telecommunications carriers for origination and termination of phone calls.
Speech-to-Text (STT)
The acoustic model and language pipeline responsible for transcribing live analog or digital telephony audio into text for the LLM to process.
T
TTS (Text-to-Speech)
The neural voice synthesis technology that converts text responses generated by the language model into high-fidelity human-sounding speech streams.
Timezone-Aware Pacing
Campaign dispatch logic that evaluates each lead's regional area code and location to ensure calls only dial within legal business hours.
Two-Sided Transcript
A synchronized, timestamped text log that separates speaker turns between the AI voice agent and the human callee for auditing and analysis.
V
Voice Activity Detection (VAD)
An audio signal algorithm that continuously distinguishes between human speech energy and ambient background acoustic noise.
Voice AI Platform
A full-stack software system orchestrating telephony carrier lines, real-time speech recognition, LLM reasoning, voice synthesis, and CRM data flow.
W
Warm Transfer
The process of conferencing a human representative into an active call as a second participant without dropping the line or requiring the customer to redial.
Webhook Call Dispositions
Real-time HTTP POST notifications dispatched immediately when a call completes, containing transcripts, recording links, duration, and captured lead attributes.
Z
Zoho CRM Integration
A native bi-directional connector that automatically pulls leads from Zoho CRM modules into calling campaigns and writes call outcomes back to contact records.