Shadowing Practice: Can This Robot Be My Personal Assistant - Learn English Speaking with Video
Creating lesson...
1
I've always dreamt of having a robot and in 2026, it feels like that dream is a little bit closer all of the time.
2
However, Very recently, it got much closer for me because on my desk, is this.
3
This is a reachy mini.
4
Now, reaching mini is a super cute robot from pollen robotics
5
and hugging face and it's not just like a product toy robot that you buy.
6
It's a straight up development platform that you could do anything you want with it.
7
But just owning the robot wasn't really the dream.
8
So I set out to connect Ricci Mini to Hermes and give myself a true robot personal assistant.
9
And because I need to justify having a robot like this, I set out on a side quest as well.
10
That side quest is to build
11
from a cute little robot like this seems way more fun than practicing it with mom and dad.
12
But before I even got here though, we needed to build Ricci because this thing comes unassembled.
13
Now Ricci wasn't an easy build but it also wasn't a difficult build.
14
It probably took an hour and 45 minutes and it could have taken much less if I was doing it by myself.
15
This and learn what all of the different component connectors were.
16
We got to learn how to use the screwdrivers and how to connect things.
17
This isn't their first electronics project, but it definitely was their most involved.
18
And when we finally got Ricci connected and turned on, there was moments of real genuine joy from them.
19
Get behind him.
20
We're private.
21
Hi, Richie!
22
Welcome to the city, marshmallow!
23
You're a winner, marshmallow!
24
So now that Ricci is built, it's time to get working on our custom software.
25
Because we have a lot of things I would like to do with this.
26
Now, the community -based apps are great, but I had bigger plans.
27
So to make Ricci a true personal assistant and claim success here, I need this robot to do a few things.
28
I need to, one, connect it to my Hermes.
29
I need to get wake word detection so that way I can talk to it directly starting out of nowhere.
30
I need to have conversation going back and forth.
31
And as far as the personal assistant tasks go, I would love for this thing to have a memory.
32
I would like for it to be able to do tool calling.
33
I would like for it to do simple coding tasks, be able to use my home assistant.
34
And I would love for this to be able to do personal assistant tasks like simply reading my calendar.
35
A lot of those are very possible with Hermes, but many of these things are going to require true custom software.
36
Now to give you some background, Ricci comes in two flavors.
37
There is the light version, which is what we have here, which requires the brains of it to be running on a computer.
38
I have a Mac mini right there that is perfect for running long running software on.
39
So I'm going to be running all of my Ricci mini software off of this computer, and Ricci is going to be communicating over wifi.
40
version of the Ricci Mini that has a built -in Raspberry Pi.
41
However, I think the light version, if you have a long -running computer, makes a ton of sense, right?
42
So what did I do?
43
Well, I started off with a fork of the existing conversation app.
44
From there, I used the prompt in the docs to really get started.
45
I then replaced all of the calls to OpenAI with calls to my Hermes, which was really simple because Hermes has a great API.
46
And this actually gave me a ton of I use my Hermes for a ton of stuff already.
47
So just simply by connecting it to Hermes, I was able to get memory via honcho,
48
tool calling of all the tools that my various profiles have set up, coding tasks so it can actually do coding on any of my machines since they're all connected via SSH.
49
So simply getting those things working was super easy,
50
but the tough stuff here was actually getting timing.
51
Basically, the Ricci software has voice activity detection.
52
It decides when speech is actually starting and ending based on the audio level that's coming in.
53
If the voice activity detection detects speech or silence,
54
it then starts to transcribe via the local SST, the speech to text that's done being whisper and it's happening on my Mac mini.
55
From there, it looks to of this robot, my kids named it because just look at the guy.
56
From then, Ricci would then acknowledge the conversation and then start conversation mode.
57
Ricci basically just listens for the next sentence and then after 1 .1 seconds of silence,
58
which by all means is a value that I am still tweaking
59
and have had to tweak based on the way kids speak to this thing, the way I speak to this thing.
60
It's taken a lot of trial and error here.
61
It then transcribes that text locally.
62
That text is basically sent to Hermes in a prompt.
63
Hermes does its thing, returns the prompt.
64
Then our software generates text to speech using Kokoro,
65
to improve upon the general speed and latency in conversation with this thing.
66
We then send that audio to the robot and the robot outputs it directly so we can hear it.
67
It then begins listening for the next prompt for me.
68
So what you can see is it's basically just a conversational prompt.
69
Now we also have some stop word detection where we say go to sleep and Rigi will stop the conversation.
70
Again, this is still things that I'm tweaking and I set out to create a whole host of where the latency is.
71
Now, one of the reasons why Hermes is less good for this is because by default, Hermes agents have a ton of additional tokens in here.
72
So the responses take a long time
73
because we're sending a lot of information in each prompt rather than quick little fast prompts.
74
It also depends, of course, which model you're using to have it respond to.
75
Now, I just have this connected via my codec subscription.
76
So it is using some chunky models rather than some highly tuned fast I will be experimenting with more in the future.
77
So did this work?
78
Well, let's start a wake listening and we're going to see
79
if all of these things work and if we can check them off our list.
80
I have Marshmallow here, we have wake word listening,
81
we have conversation, and we're gonna first give this a try with memory and see if Marshmallow can understand.
82
By the way, this is the custom UI that I've been building.
83
Hey, Marshmallow.
84
Hey, I'm listening.
85
Oh What do you know about me?
86
You're Scott.
87
Oh You like concise, direct technical progress, live checks, and family -friendly Reiki mini -apps.
88
You also prefer short robot prompts and you've got two kids who may use me for spelling practice.
89
I am listening.
90
Go to sleep.
91
Okay, I'll listen for Hey Marshmallow.
92
So, you can see that the conversation definitely works.
93
Now, the latency is certainly there.
94
It's something that I'm working on improving.
95
Okay, now let's try some of the other things we had in mind.
96
Now, I have several different profiles in Hermes that all have different abilities.
97
And the marshmallow profile is kid friendly and kids safe.
98
However, I have a default profile that has access to things like my calendar, as well as can do coding.
99
new conversation.
100
What's on my calendar for today?
101
Woohoo!
102
Jury duty all day, drop off from 8 :00 to 8 :30 a .m.
103
Release phases tomorrow from 6 :00 to 7 :00, 0 :00 PM.
104
Change the color of the Marshmallow Conversation app background to be pink.
105
Mm -hmm.
106
Now the coding tasks may take just a second here since it's going off
107
and I don't think this is even inside of the project
108
so it's probably going to have to go look for the project on my machine and then change those files.
109
Done, the Marshmallow Conversation app now has a pink background, and I verified it live at port 7860.
110
It looks good.
111
Yeah, it looks good.
112
Marshmallow, can you turn all of the lights off please?
113
Mm -hmm.
114
you you
115
all reachable lights are off so now
116
that i have the personal assistant of my dreams it's time to get working on the hard part
117
which is the spelling test now the spelling test has a
118
number of features it needs to ingest the spelling list of
119
words it needs to use them in a sentence it needs user has gotten the spelling incorrect.
120
Now, there's a ton of small little problems here and it's time to get to work.
121
So after grinding on our spelling test for quite a while, we have something that can work that allows us to even save and restore various spelling lists.
122
Now is the moment of truth.
123
We're going to be giving this spelling test to my 9 year old and see how he does.
124
J -U -M -P.
125
The word is jump.
126
It is spelled J -U -M -P, try spelling it with me i'm so mad okay
127
so the reviews were not glowing he was a harsh customer
128
and he didn't like it this is certainly not going to help him practice spelling
129
but i have some ideas on how to fix this what
130
if instead of having a back and forth that took
131
so long we pre -rendered all of the audio
132
and including pre -rendering any of the helpful text any of the words that's saying you almost got it
133
going to feel snappier back and forth.
134
And what if we also output the letters on the screen as he was saying it?
135
But most importantly, what if it actually worked?
136
Because as you saw, and honestly, what you didn't see was him spelling the words correctly over and over again, and the robot consistently getting them wrong,
137
specifically because the window for when he was supposed to son a fan of this app.
138
So because of this first initial failed test, what I actually ended up doing was setting the Ricci mini in front of a computer,
139
having that computer take the spelling test on the Ricci mini back and forth with audio and microphones.
140
And then that way, the system was able to self heal
141
and improve the latency using open AI models to run the tests check the debugging information
142
and then fix it test it reiterate until now we have an app that is i gotta say working pretty well
143
Your word is robot.
144
The teacher asked us to spell robot.
145
Um, R -O -B -O -T.
146
Nice work, you spelled it carefully.
147
Huh.
148
Your word is eat.
149
The teacher asked us to spell eat.
150
EAT.
151
Nice work, you spelled it carefully.
152
Your word is jump.
153
The class practiced the word jump today.
154
J U M P.
155
Nice work.
156
You spelled it carefully.
157
So I think this was a great success.
158
The robot is now my personal assistant.
159
It's taking spelling tests with our kids,
160
and it's only going to get better from here because I still haven't really implemented that much with the camera.
161
The next step on the docket is to use the camera to teach my son and I how to play Warhammer 40k.
162
We're going to use the camera and the LLM that this is talking to,
163
to discuss both strategy, what to do next so that way it can even identify pieces on the battlefield.
164
That is a big project and the one that I'm going to be diving into next.
165
That said, I'm a huge fan of this little robot.
166
It's a really cool open project that you can do all kinds of stuff with and I gotta say, I think I just made a new friend.
📺 Same Channel
B21
I bought Black Market AI Tokens on Chinese Amazon
Business EnglishUniversity StudentsWorking Adults
C13
STOP Running Code On Your Computer
Business EnglishUniversity StudentsWorking Adults
C14
8 Tech Choices to Lock In Before Agentmaxxing
Business EnglishUniversity StudentsWorking Adults
C13
8 New CSS, JS and HTML Features You Should Know
Business EnglishUniversity StudentsWorking Adults
B23
AI Is Stealing From You
Business EnglishDaily ConversationUniversity Students
✨ Recommend Video
B22
12 sentences
Polite To AI video
Business EnglishDaily ConversationHigh School Students
B11
40 sentences
AEF 3.3 - Harry Questions Rob
Business EnglishDaily ConversationHigh School Students
B21
190 sentences
This 25-Year-Old Built a 2-Year Moat Nobody Can Bet Against | Corgi, Nico Laqua & Emily Yuan
Business EnglishUniversity StudentsWorking Adults
B11
47 sentences
Business Partner B1 Unit 4.3.2
Business EnglishDaily ConversationHigh School Students
B21
68 sentences
GitHub Copilot now controls your command line...
Business EnglishUniversity StudentsWorking Adults
B21
169 sentences
World’s most advanced robotic warehouse (AI automation)
Business EnglishUniversity StudentsWorking Adults
B22
43 sentences
AI is changing the world, but is it also trying to kill us? @RonnyChieng investigates #DailyShow
Business EnglishDaily ConversationUniversity Students
B2251 sentences
Is Siri AI Usable for English Learning?
Business EnglishDaily ConversationUniversity Students
B21
169 sentences
Can This Vibe Coder Beat a Senior Developer?
Business EnglishUniversity StudentsWorking Adults
B21
259 sentences
G10 - Summer Guide I Techfuturist
Business EnglishInterview EnglishHigh School Students
B21
52 sentences
Robots and A1 - B2 Listening Lesson for High-Intermediate Students
Business EnglishDaily ConversationHigh School Students
C1326 sentences
6 Hermes Agent use cases I promise will change your life
Business EnglishUniversity StudentsWorking Adults
Vocabulary and speaking notes for this lesson
This video has 166 sentences and 2015 words to shadow. The speech runs for 13:48. The speaker talks at a steady 146 words per minute, a comfortable pace for shadowing. Only 81% of the words are among the 3,000 most common in English, so the vocabulary is demanding.
Key vocabulary in this video
15 less common words from the video, with pronunciation and meaning:
- robot /ˈɹoʊ.bɑt/ (noun) — An intelligent mechanical being designed to look like a human or other creature, and usually made from metal.
- spelling /ˈspɛl.ɪŋ/ (noun) — The act, practice, ability, or subject of forming words with letters, or of reading the letters of words; orthography.
- mini /ˈmɪn.i/ (adjective) — Miniature, tiny, small.
- prompt /pɹɑmpt/ (verb) — To lead (someone) toward what they should say or do.
- connect /kəˈnɛkt/ (verb) — To join (to another object): to attach, or to be intended to attach or capable of attaching, to another object.
- spell /spɛl/ (noun) — Words or a formula supposed to have magical powers.
- audio /ˈɔ.di.oʊ/ (adjective) — Focused on audible sound, as opposed to sight.
- detection /dɪˈtɛk.ʃən/ (noun) — The act or process of detecting, uncovering, or finding out, the discovery of something new, hidden, or disguised.
- carefully /ˈkɛɹf(ə)li/ (adverb) — With care; attentively, circumspectly.
- custom /ˈkʌstəm/ (noun) — Frequent repetition of the same behavior; way of behavior common to many; ordinary manner; habitual practice; method of doing, living or behaving.
- calendar /ˈkæl.ən.də/ (noun) — Any system by which time is divided into days, weeks, months, and years.
- silence /ˈsaɪ̯ləns/ (noun) — The absence of any sound.
- output /ˈaʊtpʊt/ (noun) — Production; quantity produced, created, or completed.
- default /dɪˈfɔlt/ (noun) — The condition of failing to meet an obligation.
- quest /kwɛst/ (noun) — A journey or effort in pursuit of a goal (often lengthy, ambitious, or fervent); a mission.
Phrasal verbs you will hear
- work on (verb) — To apply mental or physical effort to (something) in order to shape, form, improve, investigate, solve it, etc.
- come in /ˌkʌm ˈɪn/ (verb) — To enter.
- do with (verb) — Only used in to do with
- end up (verb) — To bring to a conclusion.
- get behind (verb) — To come to believe or support something.
- go back (verb) — To return to a place or state after having been there at a previous time.
- go off (verb) — To explode.
- light off (verb) — To light the fire on a gas boiler to turn it on.
Pronunciation to watch
The speaker uses 32 contractions and reduced forms, such as I'm, we're, wasn't. Say them the short way, as you hear them.
- The “sh” and “zh” sounds: detection /dɪˈtɛk.ʃən/, harsh /ˈhɑɹʃ/, subscription /səbˈskɹɪpʃən/, marshmallow /ˈmɑɹʃˌmɛl.oʊ/, conversational /ˌkɒnvə(ɹ)ˈseɪʃənəl/
- Long words — get the stress right: experiment /ɪkˈspɛɹ.ə.mənt/, consistently /kənˈsɪstəntli/, communicate /kəˈmjuːnɪkeɪt/, electronics /ɪˌlɛkˈtɹɑ.nɪks/, importantly /ɪmˈpɔɹ.tənt.li/
How to practise with this video
- Listen to the whole video once without speaking and note the words you do not know.
- Shadow it sentence by sentence at normal speed, repeating each one until your rhythm matches the speaker.
- Record yourself and compare with the original, paying attention to words like robot, spelling, mini.
Grammar in this video
The structures the speaker uses most, with the exact words from the video:
| Structure | In the video |
|---|---|
| Present perfect have/has + past participle — a past action that still matters now | I've always dreamt · have set · has gotten |
| Relative clauses who / which + clause — extra information about a person or thing | version, which is · Hermes, which was · kids who may |
| Passive voice be + past participle — the focus is on what happens, not who does it | is built · is spelled |
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.