AMCAT SVAR Test 2026: Sections, Tips, Sample Qs

AMCAT SVAR is a 16 to 20 minute, AI-graded spoken-English test that companies use to screen voice-process and customer-experience candidates. It’s an optional module of the broader AMCAT platform, run by SHL India. SHL acquired Aspiring Minds in 2019 and still runs the test under both names. For most students typing “SVAR test” into Google, the practical reality is simpler: you’re applying to a voice or customer-success role, and SVAR is the gate between you and an interview slot.

This article covers what SVAR actually measures in 2026, how the AI grading pipeline works, which companies use it, and a 7-day prep plan that skips the time-wasters.

What is the AMCAT SVAR test?

SVAR stands for Spoken English Voice and Accent Recognition test. It’s a tool-based assessment, run online with a webcam and microphone (the older telephone-based delivery has mostly been retired in favour of browser-based sessions). The standard test runs 16 to 20 minutes and asks around 45 questions across six sections, scored on a percentile basis per section.

A common confusion is worth clearing up before we go deeper. SVAR is not the same as Versant. Versant is Pearson’s competing spoken-English test, used by a different but overlapping set of employers. Some companies, like Concentrix, run a hybrid Versant-plus-AMCAT sequence in their hiring process. Some run pure AMCAT SVAR. A few large IT services firms have moved to their own internal voice tests. Two things stay constant across all of these: they’re all AI-graded, and they all measure roughly the same skill. If you can clear AMCAT SVAR, your prep transfers to Versant cleanly.

For full context on the wider AMCAT and its other modules, see our module-wise AMCAT syllabus guide. For test dates, fees, and eligibility specifics, the AMCAT exam dates and eligibility article is the canonical reference.

Who actually uses SVAR for hiring in 2026

AMCAT SVAR shows up most often in three hiring contexts.

Voice-process and customer-experience BPOs use it as a primary screen for English fluency before any technical or HR round. Concentrix, HGS, and Tech Mahindra BPS are the most consistent users. The full list of BPO and shared-services employers that hire through AMCAT modules is wider than this; our article on companies hiring through AMCAT, eLitmus, and CoCubes tracks the broader cohort.

Mid-size IT services firms with international clients use it for voice-adjacent roles, like client servicing and L1 support, where written English alone isn’t enough.

B2B SaaS firms have started using SVAR or its equivalent in 2025 and 2026 for premium customer-success seats. These are non-voice roles where written-plus-spoken English clarity matters because the work involves client calls, demo recordings, and Slack-based async communication with global teams.

The salary range you’re targeting depends on which of these you’re aiming at:

Role type	Typical entry CTC (2026)	What SVAR is screening for
Domestic voice / Hindi or regional process	₹2.4 to ₹3.0 LPA	Clarity and basic comprehension
International voice (US/UK clients)	₹3.0 to ₹4.5 LPA	Pronunciation, fluency, and accent neutrality under pressure
Premium customer-success / B2B SaaS	₹4.5 to ₹6.5 LPA	Conversational range and quick comprehension

The range looks wide because it is. A student clearing SVAR with the same percentile can land at very different CTCs depending on the role type, the employer’s pay bands, and how their other AMCAT modules score. For Tier-2 and Tier-3 college students without coding-strong AMCAT scores, the international-voice band is usually the realistic ceiling for the first job; the customer-success band tends to require an additional written test or technical screen.

The six sections, in order, with sample questions

Here’s what each of SVAR’s six sections actually tests, with the kind of question to expect and one specific tip per section. The question phrasings below are representative of the format, not literal past-paper items.

Section 1: Reading

You’re given printed sentences and asked to read them aloud after a beep cue. Around 12 questions. The AI is scoring pronunciation accuracy, fluency, and pace.

Sample read-aloud sentence: The supervisor mentioned that the new shift schedule will start from the first Monday of next month.

Tip: don’t try to “speak American.” The AI scores phonetic accuracy against multiple reference accents; it doesn’t reward an Indian student trying to fake a US accent. Speak the way you’d read this sentence aloud to a colleague. The system is looking for consistent rhythm and full-word articulation, not whether you say “schedule” as “skedule” or “shedule.”

Section 2: Listening (repeat)

You hear a sentence once, and you repeat it verbatim within the response window. Around 10 questions. Scored on word-level match against the original audio.

Sample audio prompt: “Please ensure all client documents are submitted before five thirty in the evening.” Your task: repeat the sentence exactly, in one go.

Tip: this section punishes paraphrasing. If you missed a word, say a placeholder syllable and continue rather than restarting from the beginning. The AI scores partial-match better than fragmented restarts.

Section 3: Listening Deductions

You listen to a 60 to 90 second conversation between two people, usually a workplace scenario, and answer multiple-choice questions about its content. The questions are factual or inferential.

Tip: most candidates lose marks here from over-listening for emotional cues. The AI grades MCQ correctness; it doesn’t care if you got “the vibe” of the conversation. Focus on extracting the factual claims as you listen, and jot them down on the rough sheet you’re allowed during the test.

Section 4: Grammar (audio)

You hear three sentence options read aloud, and you pick the grammatically correct one.

Sample format: “A: she don’t like coffee. B: she doesn’t likes coffee. C: she doesn’t like coffee.” Pick the correct option.

Tip: jot a quick A / B / C with a tick next to the correct one on your notepad. By the time the third option finishes, the first option’s nuance fades from working memory if you’re not writing it down.

Section 5: Error Identification

You hear sentences with deliberate grammatical errors and identify what’s wrong from a multiple-choice list of error types (subject-verb agreement, preposition, tense, article).

Tip: errors fall into recurring buckets. Of these, prepositions trip up Indian-English speakers most: “discuss about” (it’s “discuss”), “return back” (it’s “return”), “explain me” (it’s “explain to me”). If your default speech includes these patterns, your error-identification score will reflect it. Worth ten minutes of focused review.

Section 6: Speaking (open)

You’re given a topic prompt, you get 30 to 45 seconds of preparation time, then 45 seconds to respond.

Tip: structure beats content. The AI is looking for fluent extended speech, not insightful argument. A simple intro-body-conclusion frame (“Three things matter here. First… Second… Finally…”) consistently outscores a more nuanced single-paragraph response. Use your prep window to draft three points and the order you’ll deliver them, not to perfect your opening line.

How AI grades your audio

The AMCAT SVAR test is graded entirely by AI. No human listens to your audio in the standard scoring path. Knowing what the AI is actually measuring changes how you prepare.

The grading pipeline runs in roughly five stages:

1. Automatic Speech Recognition (ASR). Your audio is transcribed to text. The transcription model is trained on Indian-accented English, so it’s reasonably forgiving of pronunciation, but it punishes mumbling and partial words because they don’t transcribe cleanly.

2. Phonetic similarity. Your transcription is force-aligned at the phoneme level against reference pronunciations. The system computes a Phoneme Error Rate (PER), which is how far off your phoneme sequence is from the expected one. A 2026 peer-reviewed paper in Frontiers in Communication documents the exact ECAPA-TDNN + Whisper pipeline used for Indian-learner accent assessment, with reported classification accuracy in the 88 to 93% range across proficiency levels. Tools like SVAR and Pearson’s Versant use closely related pipelines.

3. Pace and fluency. The system measures words-per-minute, silence ratios, and filler-word frequency. This is where automaticity matters: speaking smoothly at 130 to 150 WPM scores higher than perfect pronunciation at 90 WPM with long pauses.

4. Content match. For sections like Reading and Listening-Repeat, the system checks how close your transcription is to the source text. For Speaking, it checks for topic-relevant keywords and coherent structure.

5. Composite scoring. The five sub-scores are weighted together into a section percentile.

Two practical implications follow.

First, the AI doesn’t care about your accent in the way a human listener might. It cares about consistency. An Indian student speaking confidently in their natural accent will outscore the same student attempting a half-formed American accent. The “facility in spoken English” frame Pearson uses for Versant, which AMCAT mirrors, explicitly rewards automaticity over vocabulary range.

Second, mumbling kills you twice. Once at the ASR stage, because the model can’t transcribe what it can’t parse. And again at the fluency stage, because mumbling reads as paused or slurred speech. Speaking clearly at conversational pace, even with grammatical mistakes, beats speaking nervously and quietly without mistakes.

If you want to dig into the actual research on how these systems work, the Frontiers paper linked at the bottom of this article is a clean, peer-reviewed entry point. It’s not light reading, but the methodology section is illuminating for anyone preparing for SVAR or Versant.

How to prepare in 7 days

This is the prep plan that works if you have one week. It assumes you have the basic English foundation already; SVAR isn’t testing whether you know English, it’s testing whether you can use it under timed conditions.

Day 1 to 2: Diagnose. Record yourself reading two paragraphs from a newspaper aloud, in a quiet room, on the same setup you’ll use for the test. Listen back. Notice three things specifically: are you mumbling at sentence ends? Are you over-pausing between phrases? Are there words you skip when reading aloud (especially “the” and “a”)? Don’t try to fix anything yet. Just notice the patterns.

Day 3 to 4: Reading and listening. Spend 30 minutes a day reading aloud. Use any English newspaper editorial or a tech blog post. Read the way you’d read it aloud to a colleague who’s listening, not slow, not performative. Pair this with 15 minutes of listening practice using BBC podcasts or Audible audiobook samples. Pause every minute and try to repeat the last sentence verbatim.

Day 5: Grammar refresher. Re-do 30 grammar MCQs from any AMCAT preparation book or platform. Focus on Indian-English error patterns: subject-verb agreement, articles (“a” versus “the”), prepositions (the discuss-about, return-back, explain-me cluster), and tense consistency. The AMCAT previous papers article has representative MCQs across these patterns.

Day 6: Speaking under timer. Pick five topics: a recent news event, your hometown, an opinion question (privatisation of education, say), a workplace scenario, and a personal experience. For each topic, get exactly 45 seconds of prep, then speak for 45 seconds. Record every attempt. Listen back the next day, not the same day; you can’t evaluate yourself accurately in the moment.

Day 7: Test conditions. Take a full mock SVAR with the timer running, in the same room and on the same equipment you’ll use for the actual test. Use the official AMCAT preparation kit linked from the SHL India site, or any free mock provided by your college placement cell. The point of Day 7 is calibrating your pacing, not improving your skills. That ship has sailed for this cycle. Calibration is genuinely useful, though, because the actual test feels faster than mocks for most candidates, and a 7-day plan that ends on calibration day instead of practice day produces meaningfully better scores.

The mistakes most students make (and what fixes them)

Five recurring patterns we see in students preparing for SVAR. Each one comes with a specific fix.

Mistake 1: Trying to sound American. The AI doesn’t grade accent neutrality the way a human listener might. It grades phonetic consistency. Speak in your natural accent at conversational pace. The “facility in spoken English” frame favours fluent Indian English over halting attempts at American intonation.

Mistake 2: Reading too slowly. Most candidates over-correct on pronunciation by reading at 80 to 90 WPM. The AI’s fluency scoring penalises pace meaningfully below conversational speed, which sits in the 110 to 130 WPM range. Practice reading aloud at the speed you’d naturally chat with a friend, then nudge slightly faster.

Mistake 3: Restarting sentences after a slip. If you stumble on a word in Reading or Listening-Repeat, don’t restart from the beginning. The ASR transcribes the whole audio; restarts read as duplicated content and tank the fluency score. Continue forward from where you stumbled.

Mistake 4: Memorising vocabulary lists. SVAR doesn’t test vocabulary range. It tests whether you can speak fluently using the vocabulary you already have. The hours you spend learning rare synonyms are better spent on read-aloud practice.

Mistake 5: Skipping the Speaking section’s prep window. The 30 to 45 seconds of preparation time before you speak is the most under-used resource in the entire test. Most students stare at the screen and wait, then begin speaking unprepared when the response window opens. Use the prep window: jot down three points, the order you’ll deliver them, and a one-line conclusion. Structured 45 seconds beats unstructured 45 seconds, every time.

A note on what comes next

SVAR is the gate to your first voice or customer-experience job. Once you’re in, the differentiator at the next career step is whether you understand the AI tools your team is increasingly using. If first-job to AI-product team is the trajectory you’re planning for, the bridge below is worth a click.