An Alliterative Alphabet for Eliana
A a
A says Ah as in ant
Ah is an adorable alpaca with albatross wings
achieving amazing aerial acrobatics in the air.
B b
B says Buh as in bike
Baby Buh is a bear in a bubble bouncing blissfully
beneath bright blue balloons on a beautiful day.
C c
C says Ca as in cake
C says Sa as in city
Ca, the cautious caterpillar
consumes a creamy concoction in a cup
while Sa, the celebrated centipede,
ceaselessly sips cider in a citrus tree.
D d
D says Duh as in did
Duh, the darling dragon knows dinner first is right
But Duh, the darling dragon devours donuts with delight.
E e
E says Eh as in bed
E says Ee as in free
The elephant named Eh is an energetic educator with elk antlers.
F f
F says Feh as in fed
Feh, the friendly fox, feasts with friends
at fancy fiestas in the forest.
G g
G says Guh as in glad
G says Juh as in large
Guh, the green gorilla, gleefully gobbles glistening grapes.
But his friend, Juh, the grumpy giraffe
gently groans and gripes when gazing at the green globs.
H h
H says Huh as in hotel
Huh, the happy hamster wears a hilarious hat while riding a horse.
I i
I says ih as in sit
Ih the interesting impala imagines icebergs.
J j
J says Juh as in jump
Juh is a jazzy jaguar jubilant about juice.
K k
K says Kuh as in kite
Kuh is a kind kea keen on cavorting with kites.
L l
L says Luh as in lot
Laughing loudly, Luh, the loquacious llama
lounges languidly in the lush, lively landscape.
M m
M says Muh as in mother
Muh is a magical monkey majestically maneuvering massive mangoes in her mystic mountain meadow.
N n
N says Nuh as in nest
Nuh the nifty narwhal wears a necktie to eat noodles.
O o
O says Oh as in slow
Oh, the obviously outrageous oyster loves her ostentatious opal.
P p
P says Puh as in put
Puh, the playful platypus presents his perfect present
on a plate: a prickly pickle.
Q q
Q says Kwuh as in quick
Kwuh, the queen of the quokkas,
quadruples her questions quarterly.
R r
R says Ruh as in race
Ruh is a ruby-powered rabbit
racing to rescue a robin from a rascally rhododendron.
S s
S says Suh as in stick
Suh, the serene swan, spreads her snow-white wings,
showcasing her stunning splendor in the sunlight.
T t
T says Tuh as in table
Tuh is a tremendous, totally tenacious tiger,
tearing through twisting, turning tunnels.
U u
U says Uh as in up
Uh is an unusual umbrella, underground with utensils,
unexpectedly united with unicorns.
V v
V says Vuh as in very
Vuh, the vivacious vaquita, vigorously vanishes on vacation.
W w
W says Wuh as in well
Wuh, is a wistful wallaby waltzing wonderfully,
wearing a wacky woolen waistcoat
while waving a wobbly wooden walking stick.
X x
X says Ks as in box
X, the excitable ax, expertly examined
the maximum exact extent of six extravagant xylophones.
Y y
Y says Yuh as in yes
Yuh, the yellowhammer,
yammers on about yummy yellow yogurt.
Z z
Z says Zuh as in zebra
Zuh, the zealous zebu, zapped zany zinnias with zest.
About this Book
I, Daniel Esplin, wrote this book to help my daughter learn her letter sounds before going to kindergarten.
AI was not as good as I had hoped at creating images of mixed-up or made-up animals.
I also couldn't get it to write decent poetry in the flowing cadence of Dr. Seuss.
But it was good at repeated use of a single letter in a sentence which should help solidify letter sounds, which was the original intent anyway.
The sounds the letters make was inspired by speakmethod.com but I used a simplified version.
Technical Details
I wrote this with vim as a single html page using a Boostrap carousel.
I wrote a couple bash scripts to access Amazon's AI models via the command line.
That way I didn't have to bother signing up for a new service and or bother with a GUI for downloading and organizing images.
One script generated sentences based on the current letter to start coming up with ideas.
The other script generated images based on a prompt.
I used some of the Fooocus prompts as a starting point.
#!/usr/bin/env bash
LETTER=$(basename "$PWD")
FILENAME="$LETTER.txt"
touch "$FILENAME"
echo >> "$FILENAME"
WORDS="Where possible prefer words that start with $LETTER or have $LETTER in them."
SENTENCE="Write 5 silly sentences with more than 15 words, each about a different animal that starts with $LETTER"
PROMPT="$SENTENCE $WORDS"
echo "$PROMPT" > text_prompt.txt
docker run --rm -ti -v ~/.aws:/root/.aws -v "$(pwd):/aws" amazon/aws-cli \
bedrock-runtime invoke-model \
--model-id arn:aws:bedrock:us-east-1:151862562105:inference-profile/us.meta.llama4-maverick-17b-instruct-v1:0 \
--body "{\"prompt\":\"$PROMPT\",\"max_gen_len\":512,\"temperature\":0.5,\"top_p\":0.9}" \
--cli-binary-format raw-in-base64-out \
--region us-east-1 \
out.txt
jq -r '.generation' out.txt >> "$FILENAME"
cat "$FILENAME"
#!/usr/bin/env bash
LETTER=$(basename "$PWD")
IDEA="happy male zebu with magical glowing horns. he stands in a field of zinnia flowers"
# So we can automatically name the files a1.jpg, a2.jpg, etc. without overwriting previous runs.
# shellcheck disable=SC2010
INDEX=$(ls -v | grep -oE '[0-9]+' | sort -n | tail -n 1)
((INDEX++))
TEXT="cute fantasy concept art of $IDEA. vibrant, cute, kawaii, studio anime, painterly, magical, fantasy art, cover art, cheerful, happy"
NEGATIVE="photographic, realistic, realism, 35mm film, dslr, cropped, frame, text, deformed, glitch, noise, noisy, off-center, deformed, cross-eyed, closed eyes, bad anatomy, ugly, disfigured, sloppy, duplicate, mutated, black and white"
echo "Generating images: $TEXT"
echo "$TEXT" > image_prompt.txt
docker run --rm -ti -v ~/.aws:/root/.aws -v "$(pwd):/aws" amazon/aws-cli \
bedrock-runtime invoke-model \
--model-id amazon.titan-image-generator-v2:0 \
--body "{\"textToImageParams\":{\"text\":\"$TEXT\",\"negativeText\":\"$NEGATIVE\"},\"taskType\":\"TEXT_IMAGE\",\"imageGenerationConfig\":{\"cfgScale\":8,\"seed\":0,\"width\":1024,\"height\":1024,\"numberOfImages\":4}}" \
--cli-binary-format raw-in-base64-out \
--region us-east-1 \
out.txt
jq -r '.images[0]' out.txt | base64 --decode > "${LETTER}${INDEX}.jpg"
((INDEX++))
jq -r '.images[1]' out.txt | base64 --decode > "${LETTER}${INDEX}.jpg"
((INDEX++))
jq -r '.images[2]' out.txt | base64 --decode > "${LETTER}${INDEX}.jpg"
((INDEX++))
jq -r '.images[3]' out.txt | base64 --decode > "${LETTER}${INDEX}.jpg"