Frequently Asked Questions

Samira is an always-on, hands-free desktop smart speaker that acts as a 24/7 dedicated language partner sitting right on your counter or desk. Instead of opening an app or scheduling expensive classes, you simply speak out loud to start practicing immediately.

We built Samira because listening and live conversation are the most intimidating parts of learning a language—and many people simply don't have anyone at home to practice with.

It brings a personalized, adaptive conversation partner into your home while completely eliminating the social anxiety of practicing out loud. It's a private, zero-judgment space to make mistakes, polish your pronunciation, and build confidence.

Samira is currently available for purchase and use in the United States, Canada, Japan, South Korea, Hong Kong, Singapore, Taiwan, the United Kingdom, Italy, Spain, Germany, and France. 

We are actively working on expanding to additional regions.

You can currently learn 6 languages: English, French, Spanish, Japanese, Korean, and Chinese.

For your native language (the language you speak to Samira for instructions, questions, etc), Samira supports: 

English, Arabic (Egypt), Arabic (Saudi Arabia), Arabic (United Arab Emirates), Auto-detect, Bengali, Chinese (Simplified), French, German, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese (Brazil), Portuguese (Portugal), Russian, Spanish (Mexico), Spanish (Spain), Turkish, Vietnamese

The Home Assistant / Alexa Problem: Standard home assistants require a hard language setting. Switch Alexa to French, and it refuses to understand your native Japanese or English. If you don't already speak fluently, the interaction breaks instantly.

The Generic ChatGPT / LLM Problem: While you can prompt an app like ChatGPT to "translate every sentence," generic voice modes frequently break character, drop system prompts, or hallucinate when processing short, single-word inputs (like "Wait," "What?", or "Say again").

The Samira Difference: Samira uses proprietary dual-parser voice processing. It dynamically toggles between two language engines simultaneously, making single-word native interruptions and alternating target/native sentence playback completely fluid out of the box.

Samira is designed in Hong Kong and manufactured in China.

Yes, the device sits on your desk or countertop, connects to your home Wi-Fi network. It does not have an ethernet port.

Yes, Samira runs on a standard wall power adapter for 24/7 availability.

No. Samira is a dedicated language companion engineered specifically for voice interaction and interactive practice. It does not function as a general-purpose Bluetooth media speaker.

No. Samira is designed exclusively as a language learning partner. It does not integrate with smart home devices (like lights, thermostats, or locks) or perform general home assistant tasks, keeping the experience distraction-free.

Samira handles everyday ambient background noise—like light music, air conditioning, or outdoor traffic—without issue. However, because its microphone array is specifically tuned to detect human speech, it will pick up other people talking nearby. For the best experience, use Samira in a space without competing background conversations.

No. It is intentionally screen-free and camera-free. It uses an ambient LED lighting system to signal status and gently remind you of daily practice.

No, Samira is not waterproof or water-resistant. It is designed for indoor use on a dry desk, countertop, or table. Please keep it away from direct water exposure or high-humidity environments.

Your package includes the Samira smart speaker hardware, a USB-C power adapter and cable, and a quick-start onboarding guide.

Yes. Every device includes a 1-year limited hardware warranty against manufacturing defects.

Yes, the speaker requires an active subscription to cover continuous cloud LLM processing, voice synthesis, and feature updates.

Write a complete answer to the most frequent questions that your customers might have, such as important product information, shipping policies, payment issues or returns.

See: Membership plans

Yes. You can instruct it to speak faster or slower, repeat phrases, or break down complex grammar nuances on demand.

Simply say "Add to my list" mid-conversation. The speaker logs the word automatically to your personal profile so you can review it later or practice them during Quickfire translation drills.

Simply say "Remove [vocab word] from my list" at any time.

Immersion Mode is designed for advanced learners. It turns off native language repetition and conducts the entire interaction strictly in your learning language unless you ask for a clarification or grammar breakdown. 

Immersion Mode can be activated at any time during daily general conversation or during roleplaying. 

Turn it on by saying something like "Immersion Mode" or "Can you speak in Japanese only". 

However, you can still ask for help in your native language. You can questions, ask for explanations, ask how to say something and  will understand you and respond natively.

No, and that is by design. Unlike traditional app software that follows rigid, pre-programmed code, Samira is powered by dynamic conversational AI. While we give Samira structured rules and exercise guardrails, it behaves like a real human tutor—adapting its phrasing, tone, and responses naturally. This means no two practice sessions are identical, giving you realistic, organic conversation rather than memorized scripts.

*Premium Feature*
You can choose Sami's voice by logging into your User Dashboard, under Language profiles. Each language can have it's own voice, and you can preview the voices beforehand.

*Premium Feature*
We offer a diverse catalog of high-quality male and female voices. You can listen to sample audio and preview every voice directly from your user dashboard.

No. Because Samira operates in a dual-language mode, every voice must be naturally bilingual—capable of speaking both your practice language and your primary language seamlessly.

For example, a localized Japanese voice may support Japanese and English flawlessly, but might not work for Japanese and Spanish. While our core neutral voices cover a wide range of global languages, certain localized or stylized voices are designed for specific pairs.

Because not every voice supports every combination, you might occasionally choose a voice that isn't compatible with your two selected languages. If a voice doesn't work for your language pair, simply head to your user dashboard and select a different voice from the catalog.

*Premium Feature*
You can customize various aspects of Sami's personality by logging into your User Dashboard.

You will have options to choose aspects such as:

• the way it provides corrections and feedback
• the vibe and energy level
• humor level
• agenda and focus
• relationship dynamic
• conversational proactivity
• topic depth

*Premium Feature*
Samira allows you to fine-tune your AI practice partner across 7 distinct personality dimensions, creating endless combinations to match your preferred learning style—whether you want a patient mentor, an energetic peer, or a formal instructor.

Additionally, your personal learning goals shape Samira’s vocabulary and tone. Samira dynamically adapts its speech register based on your conversation topic: if you talk about your career, Samira draws from professional and workplace vocabulary; if you pivot to weekend plans, it seamlessly shifts to casual social and pop-culture phrasing.

*Premium Feature*
Live Web Integration connects Samira directly to real-time internet search tools. While standard AI language models are limited by static training cutoff dates, Live Web Integration enables Samira to fetch up-to-the-minute news, weather forecasts, live sports scores, movie reviews, and trending global events—delivering fresh, real-world context directly in your practice language.

*Premium Feature*
Live Web Integration is ideal for days when you prefer passive listening over active speaking, allowing you to absorb authentic target-language audio effortlessly.

You can ask Samira to search and discuss virtually anything on the web, including:

• Morning Briefings: Get local weather forecasts, market news, or top global headlines over coffee.
• Culture & Entertainment: Hear summaries of recent movie releases, book reviews, or local event schedules.
• Sports & Current Events: Catch up on game results, league standings, or breaking news stories.
• On-Demand Curiosity: Ask about any live or recent topic, and Samira will research it instantly and discuss it in your learning language.

Without Live Web Integration, Samira operates using its base artificial intelligence model, which has a fixed knowledge cutoff date.

While your speaker will still function as a world-class conversational tutor for open roleplays, vocabulary practice, grammar help, and exam prep, it won't be able to provide answers requiring real-time internet data—such as today's weather, breaking current events, live stock prices, or recent sports scores.

The Pronunciation Practice activity is an interactive, audio-scoring exercise designed to help you speak your target language clearly and accurately.

Using advanced speech-assessment technology, Samira evaluates your spoken attempts across three distinct layers:

• Sentence-Level Fluency: Overall rhythm, pacing, and completeness.
• Word-Level Accuracy: Identifying specific words that were mispronounced, skipped, or mumbled.
• Phoneme-Level Precision: Pinpointing exact vocal sounds (such as a tricky vowel or consonant) down to the individual sound level.

After every attempt, Samira provides instant, friendly feedback in your native language, offering practical mouth-and-tongue positioning tips whenever you need to adjust a sound.

You can start the activity at any time during a conversation simply by saying, "Let's do pronunciation practice."

• Listen & Repeat: Samira will speak a practice sentence in your target language. Once read, simply repeat the sentence back out loud.
• Instant Assessment: Samira analyzes your audio and provides immediate feedback in your native language along with an overall accuracy score.
• Request Replays: If you want to hear the sentence again before making an attempt, simply say "Repeat" or "Say it again." Samira will replay the audio without penalizing your score.
• Targeted Retries: If your score falls below the target threshold, Samira will offer a specific tip, re-read the sentence for you, and give you another attempt to master it before automatically moving to the next exercise.

Pinpoint Micro-Errors: Instead of giving vague feedback like "that sounded okay," Samira identifies the exact sound or word that caused a slump in accuracy.

Physical Articulatory Tips: When you struggle with a sound that doesn't exist in your native language, Samira gives actionable physical guidance—such as where to place your tongue or how to shape your lips.

Native-Language Clarity: All feedback, scores, and correction tips are delivered in your primary native language, ensuring you understand how to fix the sound without getting confused.

Judgement-Free Confidence: You can repeat difficult phrases as many times as you need in a private, low-pressure environment before trying them in real-life conversations.

Pronunciation practice focuses on clarity and intelligibility (your sentence be understood by native speakers), while accent correction focuses on regional sound matching.

• Pronunciation (Clarity): Focuses on making sure you utter the correct phonetic sounds, stress the right syllables, and articulate words accurately so that native speakers can understand your meaning without confusion.
• Accent (Identity): An accent is the natural, unique cadence and tonal filter influenced by your native tongue. Having an accent is a completely natural part of speaking a second language and does not hinder clear communication.

Samira does not try to erase your accent or force you to sound like a native from a specific city. The goal is simply to ensure your spoken sounds are accurate enough that you are always understood effortlessly.

The listening section is often the hardest part of any language exam. To help with this, Samira's has multiple listening exercises built-in, each one specifically designed to mimic the different listening test on the exams.

Standard study tools rely on pre-recorded audio tracks that get repetitive quickly, meaning you end up memorizing the answers instead of actually building comprehension skills.

Samira solves this by adapting directly to the specific exam and tier you're targeting—whether you're preparing for JLPT N5, HSK 4, or DELE B2. It dynamically generates level-specific listening exercises that match the exact formats, question types, and pacing of your official test.

You can view all the built-in listening exercises here.

Samira includes activities for the primary official certification exams across all six of our supported target learning languages:

• Japanese: JLPT
• Mandarin Chinese: HSK
• Korean: TOPIK
• English: TOEIC & TOEFL
• Spanish: DELE
• French: DELF

Samira adapts traditional paper-and-pencil listening sections into an interactive, audio-first training experience:

• Spoken Questions & Options: In standard exams, questions or multiple-choice answer options are often printed on paper for you to read. Samira speaks the scenarios, questions, and options out loud in your target language, giving your brain double the listening exposure.
• On-Demand Repetition: If you miss a detail during a prompt, you can ask Samira to repeat the question or answer choices at any time.
• Heightened Challenge: Because you rely purely on ear-to-brain comprehension without visual or printed crutches, practicing with Samira builds faster auditory processing speed—making the actual test day feel much easier by comparison.

All the built-in Listening Exercises for each language can be found here. 

It describes how each exercise works, each exercise is numbered, and starting them is as easy as saying something like:

• Do listening exercise one.
• Let's start quick response activity.
• Short dialogue exercise.

They mimic the structure, prompt styles, and difficulty tiers of official exam sections as closely as possible within a screen-free, audio-only environment. However, they are adapted for voice interaction rather than printed test booklets.

Additionally, exercise difficulty maps to official exam design:

• High-Level Formats: If a specific activity type only exists in advanced exam tiers (e.g., Level 4 / N2 / HSK 4), selecting that exercise will start at Level 4 difficulty, even if your account is set to Level 1.
• Low-Level Formats: If an activity type only exists in foundational exam tiers (Level 3 and below), the maximum difficulty for that activity caps at Level 3, even if your profile is set to Level 5.

To keep exercise processing fast and clear, there are a few practical trade-offs:

• Standard Speech Only: Samira currently uses clear, standard accent models and does not simulate regional dialects or heavy localized accents.
• Two-Speaker Limit: Multi-person dialogues are limited to two conversational partners. Complex roundtable discussions involving 3 to 6 distinct voices are not supported.
• Audio Adaptation of Paper Formats: Formats that rely heavily on visual matching grids, complex diagrams, or long written text blocks are re-engineered into spoken multiple-choice questions or conversational prompts.

Privacy is a core design principle at Samira. We operate under a strict no-ads policy and never sell, rent, or monetize your personal data or conversation history with third-party ad networks. All user data and text logs are encrypted both in transit and at rest using enterprise-grade security standards.

We do not save or store raw audio recordings of your voice—once your spoken speech is processed, the audio is deleted immediately. Text transcripts of your conversations are stored securely to provide customer support, troubleshoot technical issues, and deliver usage reports. You can review or permanently delete your text logs at any time through your account dashboard.

No. The hardware uses localized wake-word processing and only transmits audio to the cloud when an active conversation is triggered.

No. We maintain a strict privacy policy: we never sell user data, run advertisements, or link your account to third-party ad networks like Google or Meta.

All user data, account information, and encrypted text transcripts are hosted and stored on secure cloud servers located in the United States.