library(tidyverse)
library(glue)
options(scipen = 999)

df_sound <-
  tibble(
    unit = c(
      1:19,
      seq(20, 90, by = 10)
    ),
    # 1e3, 1e4, 1e5, 1e6),
    sound = c(
      "one", "two", "three", "four", "five",
      "six", "se-ven", "eight", "nine", "ten",
      "e-le-ven", "twelve", "thir-teen", "four-teen", "fif-teen",
      "six-teen", "sev-en-teen", "eight-teen", "nine-teen",
      "twen-ty", "thir-ty", "four-ty", "fif-ty",
      "six-ty", "seven-ty", "eigh-ty", "nine-ty"
    )
  )

Base data:

I made a small dataset that has syllables for numbers 1:19, then by 10s from 20-90

sample_n(df_sound, 10) |> 
  arrange(unit)
unit sound
1 one
3 three
5 five
6 six
7 se-ven
9 nine
12 twelve
18 eight-teen
30 thir-ty
60 six-ty 
nth_digit <- function(x, n, end = n) {
  str_sub(
    string = x, 
    start = -n, 
    end = -end
  ) |> 
    as.integer()
}

nth_digit(7654321, 5)
nth_digit(7654321, 5, end = 1)

all_nums <-
  tibble(number = 1:1e6) %>% #1e6
  #sample_n(8) |>
  #filter(number %in% c(690263, 512608, 57002)) |> 
  #arrange(desc(number)) |>
  # slice(1:100) %>%
  mutate(
    d7 = nth_digit(number, 7),
    d6 = nth_digit(number, 6),
    d5 = case_when(
      between(nth_digit(number, 5, 4), 10L, 19L) ~ nth_digit(number, 5, 4),
      TRUE ~ nth_digit(number, 5) * 10L
    ),
    d4 = case_when(
      nth_digit(number, 4) == 0 ~ NA_integer_,
      between(nth_digit(number, 5, 4), 10L, 19L) ~ NA_integer_,
      TRUE ~ nth_digit(number, 4)
    ),
    d3 = case_when(
      nth_digit(number, 3) == 0 ~ NA_integer_,
      TRUE ~ nth_digit(number, 3)
    ),
    d2 = case_when(
      nth_digit(number, 2) == 0 ~ NA_integer_,
      between(nth_digit(number, 2, 1), 10L, 19L) ~ nth_digit(number, 2, 1),
      TRUE ~ nth_digit(number, 2) * 10L
    ),
    d1 = case_when(
      between(nth_digit(number, 2, 1), 10, 19) ~ NA_integer_,
      TRUE ~ nth_digit(number, 1)
    )
  ) %>%
  mutate_all(as.integer) %>%
  mutate(d7 = ifelse(is.na(d7), "", "one mill-i-on")) |> 
  print()


get_joins <-
  all_nums |> 
#  filter(number == 608035) |> 
  left_join(transmute(df_sound, d6 = unit, sound_6 = paste0(sound, "-hund-red"))) %>% 
  left_join(select(df_sound, d5 = unit, sound_5 = sound)) %>%
  left_join(select(df_sound, d4 = unit, sound_4 = sound)) %>%
  left_join(transmute(df_sound, d3 = unit, sound_3 = paste0(sound, "-hund-red"))) %>% 
  left_join(select(df_sound, d2 = unit, sound_2 = sound)) %>%
  left_join(select(df_sound, d1 = unit, sound_1 = sound)) |>
  print()

I then pull out those same numbers from my data to get a table like this (pivoted):

combine_text <-
  get_joins |> 
  #filter(number == 608035)
  mutate(across(where(is.character), replace_na, "")) |> 
  mutate(
    c_100k = paste(sound_6, sound_5, sound_4) |>  trimws(),
    c_100 = paste(sound_3, sound_2, sound_1) |>  trimws()
  )  |>  
  mutate(
    c_100k = ifelse(c_100k == "", "", paste(c_100k, "thous-and")),
    final = 
      paste(c_100k, c_100, sep = " ") |>
      str_replace_all(" {2,}", " ") |> 
      trimws(),
    syllables = str_count(final, "\\w+")
  ) |> 
  print()
combine_text |> 
  filter(number %in% c(123456, 1234)) |>
  mutate_all(as.character) |> 
  pivot_longer(-number) |> 
  pivot_wider(
    names_from = number, 
    values_from = value
  )
name 1234 123456
d7
d6 NA 1
d5 NA 20
d4 1 3
d3 2 4
d2 30 50
d1 4 6
sound_6 one-hund-red
sound_5 twen-ty
sound_4 one three
sound_3 two-hund-red four-hund-red
sound_2 thir-ty fif-ty
sound_1 four six
c_100k one thous-and one-hund-red twen-ty three thous-and
c_100 two-hund-red thir-ty four four-hund-red fif-ty six
final one thous-and two-hund-red thir-ty four one-hund-red twen-ty three thous-and four-hund-red fif-ty six
syllables 9 14

What I learned

(
  combine_text |>  
    select(
      number:d1, #t_100k, t_100, 
      final, syllables
    ) |> 
    ggplot(aes(syllables)) +
    geom_histogram(binwidth = 1, color = "black") +
    #theme(aspect.ratio = 1/1.6) +
    scale_y_continuous(
      labels = scales::number_format(scale = 1/1000, suffix = "K")
    ) +
    labs(
      title = "There are 330K numbers between 1 and 1M that have 14 syllables",
      y = "# of numbers"
    )
) |> plotly::ggplotly()

how long would it take?

# syllables per second, my friend was able to count to 100 in 60 seconds
sps <- 
  ceiling(
    sum(combine_text$syllables[1:100]) / 60
  )

#totals
tibble(n = 10^(2:6)) |> 
  mutate(
    mins = map_dbl(
      .x = n, 
      .f = 
        ~(sum(combine_text$syllables[1:.x]) / sps / 60) |> 
        ceiling()
    ),
    units = case_when(
      mins > 60 * 24 ~ "days",
      mins > 60 ~ "hours",
      TRUE ~ "mins"
    ),
    value =
      round(
        mins / case_when(
          units == "days" ~  60 * 24,
          units == "hours" ~ 60,
          TRUE ~ 1
        ),
        1
      ),
    n = scales::number(n, big.mark = ","),
 ) |> 
 transmute(
   counting = glue("1 to {n} would take {value} {units}")
 )
counting
1 to 100 would take 1 mins
1 to 1,000 would take 19 mins
1 to 10,000 would take 4.6 hours
1 to 100,000 would take 2.4 days
1 to 1,000,000 would take 30.1 days

If a person had 1 min rest for every 5 min of speaking for 8 hours a day, it would take an even longer time for them to finish

tibble(
  total_mins = sum(combine_text$syllables) / sps / 60, # total min (43K min)
  with_breathing = total_mins * 19/17, # 1 breath every 17 syllables (48K min)
  with_rests = with_breathing * (6/5), # 6 min for every 5 (58K min)
  n_hours = with_rests / 60, # convert to hours (968 hours)
  n_8_hour_days = n_hours / 8, # max 8 hours per day (122.8 days)
  n_business_days = n_8_hour_days * 7/5   # weekends off (170.8 days)
) |> 
  gather("metric", "time") |> 
  mutate(
    metric = 
      metric |> 
      str_replace_all("_", " ") |> 
      str_replace_all("N ", "# ") |> 
      str_to_title(),
    time = scales::number(time, big.mark = ",")
  )
metric time
Total Mins 43,327
With Breathing 48,424
With Rests 58,109
N Hours 968
N 8 Hour Days 121
N Business Days 169