Initial commit of TTS training and Berkshire Meeting 1994 sounds.

.gitignore

@@ -0,0 +1,2 @@
+rnn.raw
+temp.raw

voice-training-sets/berkshire-meetings/1994/README.md

@@ -0,0 +1,39 @@
+
+## Index of Berkshire Annual Meeting Transcripts
+
+From Warren Buffett archive at CNBC.
+
+* https://buffett.cnbc.com/video/1994/04/25/morning-session---1994-berkshire-hathaway-annual-meeting.html
+
+# HOWTO To Create Training Set of GOAT Warren Buffett for Mozilla TTS
+
+Following https://stackoverflow.com/a/66307612
+
+Use the `LiveRecorder` Firefox plugin to record from Youtube.
+
+Convert from webm to audio-only ogg.
+```
+ffmpeg -i berkshire-1994.webm -vn -acodec copy ./berkshire-1994-00:00_10:00.ogg
+```
+
+Split by silence. This automated splitting of files is the killer app for command-line audio tools, imo.
+
+```
+sox ./berkshire-1994-00:00_10:00.ogg berkshire-1994-00:00_10:00-.wav silence 1 0.2 0.5% 1 0.2 0.5% : newfile : restart
+```
+
+Play with trimming (start and duration)
+then when you narrow down the clip, you can use `sox` to trim.
+This is where GUIs like Audacity are still better.
+
+```
+play ./berkshire-1994-00:00_10:00-149.wav trim 0:00 0:02.5
+sox ./berkshire-1994-00:00_10:00-149.wav ./berkshire-1994-00:00_10:00-149a.wav trim 0:00 0:02.5
+
+```
+
+https://stackoverflow.com/questions/9667081/how-do-you-trim-the-audio-files-end-using-sox
+
+https://unix.stackexchange.com/questions/381890/play-audio-file-from-a-certain-time-step-in-terminal
+
+

voice-training-sets/berkshire-meetings/1994/metadata.csv

@@ -0,0 +1,100 @@
+berkshire-1994-00:00_10:00-001|Put this over here.
+berkshire-1994-00:00_10:00-003|Am I live yet? Yeah.
+berkshire-1994-00:00_10:00-004|Morning
+berkshire-1994-00:00_10:00-005|We were a little worried today because we weren’t sure
+berkshire-1994-00:00_10:00-006|from the reservations whether we can handle everybody but it looks to me like
+berkshire-1994-00:00_10:00-007|there
+berkshire-1994-00:00_10:00-009|we're going to have to uh
+berkshire-1994-00:00_10:00-010|find a different spot because it looks to me like we’re up about 600 this year from last year
+berkshire-1994-00:00_10:00-011|be on the safe side we will seek out a larger spot. Now, there
+berkshire-1994-00:00_10:00-012|certain implications to that because
+berkshire-1994-00:00_10:00-013|some of the more experienced of you know, a few years ago we were
+berkshire-1994-00:00_10:00-014|we were holding this meeting at the Joslyn
+berkshire-1994-00:00_10:00-015|Museum which is a
+berkshire-1994-00:00_10:00-016|culture
+berkshire-1994-00:00_10:00-017|we've now, of course, moved to
+berkshire-1994-00:00_10:00-019|an old vaudeville theater. And
+berkshire-1994-00:00_10:00-020|the only place in town that can hold us
+berkshire-1994-00:00_10:00-021|year, I think, is the Ak-Sar-Ben Coliseum where they have keno
+berkshire-1994-00:00_10:00-022|we are sliding down the cultural
+berkshire-1994-00:00_10:00-022a|chain just as 
+berkshire-1994-00:00_10:00-023|I have some rather distressing news to report
+berkshire-1994-00:00_10:00-024|There are always a few people who vote against
+berkshire-1994-00:00_10:00-025|everyone
+berkshire-1994-00:00_10:00-026|on the slate for directors, and there are maybe a dozen or so people do that. Then there are others who single shot it, and they pick out people
+berkshire-1994-00:00_10:00-027|to vote against
+berkshire-1994-00:00_10:00-028|and
+berkshire-1994-00:00_10:00-029|come as news to Charlie. I haven't told him yet but
+berkshire-1994-00:00_10:00-030|he is the only one among our candidates for directors who received no
+berkshire-1994-00:00_10:00-031a|negative votes
+berkshire-1994-00:00_10:00-031b|this year
+berkshire-1994-00:00_10:00-031c|when you lose out the title of Miss Congeniality to Charlie, you know you're in trouble
+berkshire-1994-00:00_10:00-032|I'd like to uh
+berkshire-1994-00:00_10:00-033.wav|I'd like to tell you a little bit how we'll run this. We will we will
+berkshire-1994-00:00_10:00-034|We will have the business meeting
+berkshire-1994-00:00_10:00-035|in a hurry with
+berkshire-1994-00:00_10:00-36a|of all of you, and then we will uh
+berkshire-1994-00:00_10:00-037|introduce our managers uh
+berkshire-1994-00:00_10:00-038|who are here
+berkshire-1994-00:00_10:00-039|and then we will have a uh
+berkshire-1994-00:00_10:00-040|Q and A uh
+berkshire-1994-00:00_10:00-041|period
+berkshire-1994-00:00_10:00-042|we will run that until
+berkshire-1994-00:00_10:00-043|twelve o' clock
+berkshire-1994-00:00_10:00-044|break
+berkshire-1994-00:00_10:00-045|and then at 12:15 if the hardcore want to stick around we will have another hour, so until about 1:15
+berkshire-1994-00:00_10:00-046|you're free to leave at any time, of course. I've pointed out in the past that 
+berkshire-1994-00:00_10:00-047a|better form if you leave while Charlie is talking rather than when I'm talking but
+berkshire-1994-00:00_10:00-047b|Feel free anytime
+berkshire-1994-00:00_10:00-047c|If you're panicked and you're
+berkshire-1994-00:00_10:00-047d|and you're worried about
+berkshire-1994-00:00_10:00-049|by leaving
+berkshire-1994-00:00_10:00-050a|you will be able to leave at 
+berkshire-1994-00:00_10:00-051a|we will have
+berkshire-1994-00:00_10:00-052|buses out front
+berkshire-1994-00:00_10:00-053|you to uh
+berkshire-1994-00:00_10:00-054|the hotels or the uh
+berkshire-1994-00:00_10:00-058|place in town in which we have a commercial interest
+berkshire-1994-00:00_10:00-059a|we encourage you staying around on that basis
+berkshire-1994-00:00_10:00-061|Let's have the let's get the business of the meeting out of the way. Then we can get on to more interesting things
+berkshire-1994-00:00_10:00-062|I will first uh
+berkshire-1994-00:00_10:00-063|introduce the Berkshire Hathaway directors that are
+berkshire-1994-00:00_10:00-065|present in addition to myself
+berkshire-1994-00:00_10:00-066a|there's Charlie, who is vice chairman of Berkshire
+berkshire-1994-00:00_10:00-067|and if the rest of you will stand
+berkshire-1994-00:00_10:00-069|Susan T. Buffett
+berkshire-1994-00:00_10:00-070|Howard Buffett
+berkshire-1994-00:00_10:00-071|Malcolm Chace the Third
+berkshire-1994-00:00_10:00-072|and Walter Scott Junior
+berkshire-1994-00:00_10:00-073a|Also with us today are partners in the firm of Deloitte and Touche, our auditors, Mr. Ron Burgess and Mr. Craig Christiansen
+berkshire-1994-00:00_10:00-074|They are available to respond to appropriate questions you might concerning their firm's audit of the accounts of Berkshire
+berkshire-1994-00:00_10:00-075|Forrest Krutter, secretary of Berkshire. He will make a written record of the proceedings
+berkshire-1994-00:00_10:00-076|Robert M. Fitzsimmons has been appointed inspector of election
+berkshire-1994-00:00_10:00-077|meeting. He will certify to the count of votes
+berkshire-1994-00:00_10:00-078|cast in the election for directors.
+berkshire-1994-00:00_10:00-079|The named proxy holders for this meeting are Walter Scott Jr. and Marc D. Hamburg
+berkshire-1994-00:00_10:00-080|cards have been returned through last Friday
+berkshire-1994-00:00_10:00-081|presenting 1,035,680 Berkshire shares to be voted by the proxy holders as indicated on the cards.
+berkshire-1994-00:00_10:00-082|number of shares represents a quorum and we will therefore
+berkshire-1994-00:00_10:00-083|directly proceed with the meeting
+berkshire-1994-00:00_10:00-084|conduct the business of the meeting and then adjourn to the formal meeting
+berkshire-1994-00:00_10:00-086|of the minutes of the last meeting of shareholders
+berkshire-1994-00:00_10:00-087|I recognize Mr. Walter Scott Jr. who will place a motion before the meeting
+berkshire-1994-00:00_10:00-089|Do I hear a second?
+berkshire-1994-00:00_10:00-090a|Motion has been moved and seconded. Are there any comments or questions? Hearing none, we will vote on the motion by voice
+berkshire-1994-00:00_10:00-090b|All those in favor say aye.
+berkshire-1994-00:00_10:00-106|shareholder is present who wishes to withdraw a proxy previously sent in and vote in person, he or she may do so 
+berkshire-1994-00:00_10:00-107|Also if any shareholder that's present has not turned in a proxy and desires a ballot in order to vote in person, you may do so 
+berkshire-1994-00:00_10:00-108|you wish to do this, please identify yourself ot meeting officials in the aisles who will furnish a ballot to you
+berkshire-1994-00:00_10:00-109|persons desiring ballots please identify themselves so that we may distribute them 
+berkshire-1994-00:00_10:00-110|I now recognize Mr. Walter Scott Jr. to place a motion before the meeting with respect election of directors.
+berkshire-1994-00:00_10:00-117|Are there any other nominations?
+berkshire-1994-00:00_10:00-118|there any discussion?
+berkshire-1994-00:00_10:00-119|Motion and nominations are ready to be acted upon.
+berkshire-1994-00:00_10:00-120|there are any shareholders voting in person they
+berkshire-1994-00:00_10:00-121|they should now mark their ballots and allow the ballots to be delivered to the inspector of elections
+berkshire-1994-00:00_10:00-144|Now, I'd like to introduce
+berkshire-1994-00:00_10:00-145|some of the people who make this place work
+berkshire-1994-00:00_10:00-148|and if you would
+berkshire-1994-00:00_10:00-149a|hold your applause until the end because there are quite a number of our managers
+berkshire-1994-00:00_10:00-150|I'm not sure which ones for sure are here, some of them may be out

voice-training-sets/berkshire-meetings/1994/noise_cleanup.py

@@ -0,0 +1,49 @@
+from pathlib import Path
+
+import os
+import subprocess
+import soundfile as sf
+import pyloudnorm as pyln
+import sys
+
+homedir = os.environ['HOME']
+
+src = sys.argv[1]
+rnn = f"{homedir}/src/rnnoise/examples/rnnoise_demo"
+
+paths = Path(src).glob("**/*.wav")
+
+for filepath in paths:
+    target_filepath=Path(str(filepath).replace("original", "converted"))
+    target_dir=os.path.dirname(target_filepath)
+
+    if (str(filepath) == str(target_filepath)):
+        raise ValueError("Source and target path are identical: " + str(target_filepath))
+
+    print("From: " + str(filepath))
+    print("To: " + str(target_filepath))
+
+    # Stereo to Mono; upsample to 48000Hz
+    subprocess.run(["sox", filepath, "48k.wav", "remix", "-", "rate", "48000"])
+    subprocess.run(["sox", "48k.wav", "-c", "1", "-r", "48000", "-b", "16", "-e", "signed-integer", "-t", "raw", "temp.raw"]) # convert wav to raw
+    subprocess.run([rnn, "temp.raw", "rnn.raw"]) # apply rnnoise
+    subprocess.run(["sox", "-r", "48k", "-b", "16", "-e", "signed-integer", "rnn.raw", "-t", "wav", "rnn.wav"]) # convert raw back to wav
+
+    subprocess.run(["mkdir", "-p", str(target_dir)])
+    subprocess.run(["sox", "rnn.wav", str(target_filepath), "remix", "-", "highpass", "100", "lowpass", "7000", "rate", "22050"]) # apply high/low pass filter and change sr to 22050Hz
+
+    data, rate = sf.read(target_filepath)
+
+    # peak normalize audio to -1 dB
+    peak_normalized_audio = pyln.normalize.peak(data, -1.0)
+
+    # measure the loudness first
+    meter = pyln.Meter(rate) # create BS.1770 meter
+    loudness = meter.integrated_loudness(data)
+
+    # loudness normalize audio to -25 dB LUFS
+    loudness_normalized_audio = pyln.normalize.loudness(data, loudness, -25.0)
+
+    sf.write(target_filepath, data=loudness_normalized_audio, samplerate=22050)
+
+    print("")