Shadowing Practice: Gemma 4 Just Released - Whats In? In 2026! - Learn English Speaking with Video

Choose a practice mode

Each mode opens a full-screen practice room.

28 sentences
If sentences are too short or too long, click Edit to adjust them.
1
A 2 billion parameter AI model, running on a Raspberry Pi with 1 .5 GB of RAM, and it can see, hear and think in 140 languages,
2
this is Gemma 4.
3
Google just open sourced their most capable small model family ever.
4
Gemma 4 is built on the same architecture as Gemini 3, but packaged into size that actually fit on your hardware.
5
There are four variants.
6
The E2B model uses around 2 billion parameters at inference.
7
The E4B steps up to 4 billion.
8
Then there's a 26 billion mixture of experts model and a 31 billion dense model.
9
All of them are released under Apache 2 .0, which means fully commercial use, no royalties, no strings attached.
10
What makes Gemma 4 different from previous small models is what it can actually do.
11
All four variants natively process images and video, with built -in OCR and chart understanding.
12
The two smaller models also handle audio input, meaning speech recognition runs directly on device.
13
And every model works across over 140 languages out of the box.
14
Context windows go up to 128k tokens on the smaller models and 256k on the larger ones.
15
That's a huge amount of context for something that fits on a single GPU.
16
But the real story here is a Gentic AI on the edge.
17
Jammer 4 has native function calling and structured JSON output built in, no fine -tuning needed.
18
You can build multi -step AI agents that plan, execute and adapt, all running locally on a phone, laptop or even a Raspberry Pi.
19
No cloud dependency, no API costs, no data leaving your device.
20
It runs on iOS, Android, Windows, Linux and even in the browser through WebGPU.
21
And the performance backs it up.
22
The 31 billion parameter model ranked number 3 on the ARENA AI leaderboard, among all open models worldwide.
23
The Gemma model family has already been downloaded over 400 million times with over 100 ,000 community variants.
24
Google just made frontier -level AI something you can download, run at home and build real products with.
25
No cluster required.
26
One GPU.
27
Apache 2.
28
This changes everything for Edge AI.

Download App

Everything you need to speak fluently

AI PronunciationScore every sentence
IPA PracticeMaster every sound
VocabularyBuild your word bank
Vocab GameLearn while playing
FOR YOU

Recommend Video

Vocabulary and speaking notes for this lesson

This video has 28 sentences and 336 words to shadow. The speech runs for 2:13. The speaker talks at a natural 151 words per minute, close to everyday conversation. Only 74% of the words are among the 3,000 most common in English, so the vocabulary is demanding.

Key vocabulary in this video

15 words from the video worth learning, with pronunciation and meaning:

WordPronunciationMeaning
variant noun/ˈvæɹi.ənt/Something that is slightly different from a type or norm.
parameter noun/pəˈɹæm.ə.tɚ/A value kept constant during an experiment, equation, calculation, or similar, but varied over other versions of the experiment, equation, calculation, etc.
audio adjective/ˈɔ.di.oʊ/Focused on audible sound, as opposed to sight.
chart noun/t͡ʃɑɹt/A map illustrating the geography of a specific phenomenon.
recognition noun/ˌɹɛkəɡˈnɪʃ(ə)n/The act of recognizing or the condition of being recognized (matching a current observation with a memory of a prior observation of the same entity).
cloud noun/ˈklaʊ̯d/A visible mass of water droplets suspended in the air.
package noun/ˈpæk.ɪd͡ʒ/Something which is packed, a parcel, a box, an envelope.
rank noun/ˈɹæŋk/A row of people or things organized in a grid pattern, often soldiers.
output noun/ˈaʊtpʊt/Production; quantity produced, created, or completed.
architecture noun/ˈɑː.kɪˌtɛk.t͡ʃə/The art and science of designing and managing the construction of buildings and other structures, particularly if they are well proportioned and decorated.
string noun/stɹɪŋ/A long, thin and flexible structure made from threads twisted together.
input noun/ˈɪnpʊt/The act or process of putting in; infusion.
hardware noun/ˈhɑɹdˌwɛɹ/Fixtures, equipment, fasteners, tools, and devices used for general-purpose construction and repair of a structure or object. Also such equipment as sold as…
tune noun/tjuːn/A melody.
laptop noun/ˈlæp.tɑp/A laptop computer.

Pronunciation to watch

  • Long words — get the stress right: parameter /pəˈɹæm.ə.tɚ/, recognition /ˌɹɛkəɡˈnɪʃ(ə)n/, architecture /ˈɑː.kɪˌtɛk.t͡ʃə/, dependency /dɪˈpɛndənsi/

How to practise with this video

  1. Listen to the whole video once without speaking and note the words you do not know.
  2. Start at 0.75× speed, shadow it sentence by sentence, then go back to normal speed once it feels easy.
  3. Record yourself and compare with the original, paying attention to words like variant, parameter, audio.

What is the Shadowing Technique?

Shadowing is a science-backed language learning technique originally developed for professional interpreter training and popularized by polyglot Dr. Alexander Arguelles. The method is simple but powerful: you listen to native English audio and immediately repeat it out loud — like a shadow following the speaker with just a 1–2 second delay. Unlike passive listening or grammar drills, shadowing forces your brain and mouth muscles to simultaneously process and reproduce real speech patterns. Research shows it significantly improves pronunciation accuracy, intonation, rhythm, connected speech, listening comprehension, and speaking fluency — making it one of the most effective methods for IELTS Speaking preparation and real-world English communication.

Shadowing technique: read the full step-by-step guide →