A quick R script to convert a 538 chat to an mp3 file using voices in MacOS. Requires ffmpeg, you would also need to update some of the file locations.

538_pseudo_cast.R

library('rvest')
library('dplyr')
library('magrittr')

setwd('~/src/538/')

# this is the blog post we are interested in
#URL <- "https://fivethirtyeight.com/features/who-killed-the-health-care-bill/"
#URL <- "https://fivethirtyeight.com/features/which-gop-senators-might-resurrect-the-health-care-bill/"
URL <- "https://fivethirtyeight.com/features/how-has-the-radical-right-evolved-under-trump/"

episode <- last(strsplit(URL,'/')[[1]])

#######################################
# get the blog post
#######################################
blog_538 <- read_html(URL)

#######################################
# get politics talk
#######################################
politics_talk <- blog_538 %>% html_nodes("#content") %>% html_nodes("p") %>% html_text() 

#######################################
# first index is date and time
# second index is "A FiveThirtyEightChat"
# third index is "Filed under Health Care"
# fourth is the intro, it may actually be multiple lines, neh
#######################################

#######################################
# we will terminate on "Filed under"
#######################################

people_names <- politics_talk[sapply(  politics_talk,  function(x) grepl("[)]:",x) )]
people_names <- sapply(  politics_talk,  function(x) strsplit(x,"):")[[1]][1] )

people_names <- people_names[sapply( people_names, function(x) grepl("^[a-zA-Z]([a-zA-Z]|[.]|[-]|[_]|[:])* [(]",x,ignore.case=TRUE))]
names(people_names) <- NULL

#######################################
# get short names
#######################################
short_names = sapply( people_names, function(x) strsplit(x," ")[[1]][1] )

#######################################
# place holder for reader
#######################################
politics_talk_reader <- rep(NA, length=length(politics_talk))
 
for( short_name in short_names) {
  politics_talk_reader[ grepl( sprintf("^%s",short_name),politics_talk) ] <- short_name
} 

#######################################
# identify blanks  
#######################################
politics_talk_reader[ sapply(politics_talk, nchar) == 0 ] <- ""

#######################################
# remove informational lines at the end 
#######################################
wait_to_na = FALSE
for( i in length(politics_talk):1) {
  print(i)
  if( tolower("filed under") == tolower(politics_talk[[i]]) ) {
    print('found')
    wait_to_na = TRUE
  }
  if( wait_to_na ) {
    if( nchar(politics_talk[[i]]) == 0 )  {
      break
    }
  }
  
  politics_talk_reader[i] <- ""
}




#######################################
# pull down names
#######################################
cur_name = ""
for( i in 1:length(politics_talk)) {
 
  if(is.na(politics_talk_reader[i])) {
    politics_talk_reader[i] <- cur_name
  } else if ( nchar(politics_talk_reader[i]) > 0 ) {
    cur_name = politics_talk_reader[i]
  }
  
}

politicas_talk_reader[4] <- 'narrator'

politics_talk_df <- data.frame( short_names = politics_talk_reader, stringsAsFactors = FALSE)
print(politics_talk_df)

#######################################
# assign voices to people
#######################################
mac_names <- c(
  'Alex',                #en_US    # Most people recognize me by my voice.
  'Daniel',              #en_GB    # Hello, my name is Daniel. I am a British-English voice.
  'Fiona',               #en-scotland # Hello, my name is Fiona. I am a Scottish-English voice.
  'Fred',                #en_US    # I sure like being inside this fancy computer
  'Karen',               #en_AU    # Hello, my name is Karen. I am an Australian-English voice.
  'Moira',               #en_IE    # Hello, my name is Moira. I am an Irish-English voice.
  'Rishi',               #en_IN    # Hello, my name is Rishi. I am an Indian-English voice.
  'Samantha',            #en_US    # Hello, my name is Samantha. I am an American-English voice.
  'Tessa',               #en_ZA    # Hello, my name is Tessa. I am a South African-English voice.
  'Veena',               #en_IN    # Hello, my name is Veena. I am an Indian-English voice.
  'Victoria'             #en_US    # Isn't it nice to have a computer that will talk to you?
)

# create data frame
name_df <- data.frame(
 full_names=c(sprintf("%s)",names(short_names)),'narrator'),
 short_names=c(short_names,'narrator'),
 stringsAsFactors = FALSE)

name_df$mac_names <- sample( mac_names, size=NROW(name_df), replace = FALSE )
rownames(name_df) <-NULL


# record to aiff with play
reader_df <- left_join( politics_talk_df, name_df, by="short_names")

success <- c()
for( i in 1:NROW(reader_df)){
  if( !is.na(politics_talk[[i]]) & !is.na(reader_df$mac_names[i]) ) {
    # write audio
    con <- file(sprintf('tmp/tmp%d.txt',i),'w')
    writeLines(politics_talk[[i]], con=con)
    close(con)

    success <- c(success,i)
    
    system(sprintf("say -v %s -f tmp/tmp%d.txt -o tmp/out_%d.aiff", reader_df$mac_names[i], i, i))
    unlink(sprintf('tmp/tmp%d.txt',i))
  }
}


con <- file(sprintf('tmp/all_files.txt'),'w')
writeLines(sprintf("file out_%d.aiff",success), con=con)
close(con)

system( "ffmpeg -f concat -i tmp/all_files.txt -c copy output.aiff")
system( sprintf("ffmpeg -i output.aiff %s.mp3 -y", episode))
unlink('output.aiff')