Merge branch 'main' of http://199.170.132.91:3000/cryptogoth/value-investors-tts into main

.gitignore

@@ -1,2 +1,4 @@
 rnn.raw
 temp.raw
+3-wavs
+*.wav

annual-letters/bpl-1965.txt

@@ -0,0 +1,46 @@
+Buffett Partnership Letters 1957 to 1970
+www.csinvesting.wordpress.com studying/teaching/investing Page 67
+BUFFETT PARTNERSHIP, LTD.
+810 KIEWIT PLAZA
+OMAHA 31, NEBRASKA
+January 18, 1965
+Our Performance in 1964
+Although we had an overall gain of $4,846,312.37 in 1964, it was not one of our better years as judged by our
+fundamental yardstick, the Dow-Jones Industrial Average (hereinafter called the "Dow"). The overall result for
+BPL was plus 27.8% compared to an overall plus 18.7% for the Dow. The overall result for limited partners was
+plus 22.3%. Both the advantage of 9.1 percentage points on a partnership basis and 3.6 points by the limited
+partners were the poorest since 1959, which was a year of roughly comparable gains for the Dow.
+Nevertheless, I am not depressed. It was a strong year for the general market, and it is always tougher for us to
+outshine the Dow in such a year. We are certain to have years when the Dow gives us a drubbing and, in some
+respects, I feel rather fortunate that 1964 wasn't the year. Because of the problems that galloping markets pose
+for us, a Dow repeat in 1965 of 1964 results would make it most difficult for us to match its performance, let
+alone surpass it by a decent margin.
+To bring the record up to date, the following summarizes the year-by-year performance of the Dow, the
+performance of the Partnership before allocation to the general partner, and the limited partner's results:
+Year Overall Results From
+Dow (1)
+Partnership Results (2) Limited Partners’
+Results (3)
+1957 -8.4% 10.4% 9.3%
+1958 38.5% 40.9% 32.2%
+1959 20.0% 25.9% 20.9%
+1960 -6.2% 22.8% 18.6%
+1961 22.4% 45.9% 35.9%
+1962 -7.6% 13.9% 11.9%
+1963 20.6% 38.7% 30.5%
+1964 18.7% 27.8% 22.3%
+(1) Based on yearly changes in the value of the Dow plus dividends that would have been received through
+ownership of the Dow during that year. The table includes all complete years of partnership activity.
+(2) For 1957-61 consists of combined results of all predecessor limited partnerships operating throughout
+the entire year after all expenses, but before distributions to partners or allocations to the general
+partner.
+(3) For 1957-61 computed on the basis of the preceding column of partnership results allowing for
+allocation to the general partner based upon the present partnership agreement, but before monthly
+withdrawals by limited partners.
+On a cumulative or compounded basis, the results are:
+Year Overall Results From
+Dow
+Partnership Results Limited Partners’
+Results
+1957 -8.4% 10.4% 9.3%
+1957 – 58 26.9% 55.6% 44.5%

bpl/1957/1-in/bpl-1957-letter.txt

@@ -0,0 +1,100 @@
+WARREN E. BUFFETT
+
+5202 Underwood Ave. Omaha, Nebraska
+
+SECOND ANNUAL LETTER TO LIMITED PARTNERS
+
+The General Stock Market Picture in 1957
+
+In last year's letter to partners, I said the following:
+
+My view of the general market level is that it is priced above intrinsic value. This view relates to blue-chip
+securities. This view, if accurate, carries with it the possibility of a substantial decline in all stock prices, both
+undervalued and otherwise. In any event I think the probability is very slight that current market levels will be
+thought of as cheap five years from now. Even a full-scale bear market, however, should not hurt the market
+value of our work-outs substantially.
+
+If the general market were to return to an undervalued status our capital might be employed exclusively in
+general issues and perhaps some borrowed money would be used in this operation at that time. Conversely, if
+the market should go considerably higher our policy will be to reduce our general issues as profits present
+themselves and increase the work-out portfolio.
+
+All of the above is not intended to imply that market analysis is foremost in my mind. Primary attention is given
+at all times to the detection of substantially undervalued securities.
+
+The past year witnessed a moderate decline in stock prices. I stress the word "moderate" since casual reading of
+the press or conversing with those who have had only recent experience with stocks would tend to create an
+impression of a much greater decline. Actually, it appears to me that the decline in stock prices has been
+considerably less than the decline in corporate earning power under present business conditions. This means that
+the public is still very bullish on blue chip stocks and the general economic picture. I make no attempt to
+forecast either business or the stock market; the above is simply intended to dispel any notions that stocks have
+suffered any drastic decline or that the general market, is at a low level. I still consider the general market to be
+priced on the high side based on long term investment value.
+
+Our Activities in 1957
+
+The market decline has created greater opportunity among undervalued situations so that, generally, our
+portfolio is heavier in undervalued situations relative to work-outs than it was last year. Perhaps an explanation
+of the term "work-out" is in order. A work-out is an investment which is dependent on a specific corporate
+action for its profit rather than a general advance in the price of the stock as in the case of undervalued
+situations. Work-outs come about through: sales, mergers, liquidations, tenders, etc. In each case, the risk is that
+something will upset the applecart and cause the abandonment of the planned action, not that the economic
+picture will deteriorate and stocks decline generally. At the end of 1956, we had a ratio of about 70-30 between
+general issues and work-outs. Now it is about 85-15.
+
+During the past year we have taken positions in two situations which have reached a size where we may expect
+to take some part in corporate decisions. One of these positions accounts for between 10% and 20% of the
+portfolio of the various partnerships and the other accounts for about 5%. Both of these will probably take in the
+neighborhood of three to five years of work but they presently appear to have potential for a high average annual
+rate of return with a minimum of risk. While not in the classification of work-outs, they have very little
+dependence on the general action of the stock market. Should the general market have a substantial rise, of
+course, I would expect this section of our portfolio to lag behind the action of the market.
+
+Results for 1957
+
+In 1957 the three partnerships which we formed in 1956 did substantially better than the general market. At the
+beginning of the year, the Dow-Jones Industrials stood at 499 and at the end of the year it was at 435 for a loss
+of 64 points. If one had owned the Averages, he would have received 22 points in dividends reducing the overall
+loss to 42 points or 8.470% for the year. This loss is roughly equivalent to what would have been achieved by
+investing in most investment funds and, to my knowledge, no investment fund invested in stocks showed a gain
+for the year.
+
+All three of the 1956 partnerships showed a gain during the year amounting to about 6.2%, 7.8% and 25% on
+yearend 1956 net worth. Naturally a question is created as to the vastly superior performance of the last
+partnership, particularly in the mind of the partners of the first two. This performance emphasizes the
+importance of luck in the short run, particularly in regard to when funds are received. The third partnership was
+started the latest in 1956 when the market was at a lower level and when several securities were particularly
+attractive. Because of the availability of funds, large positions were taken in these issues. Whereas the two
+partnerships formed earlier were already substantially invested so that they could only take relatively small
+positions in these issues.
+
+Basically, all partnerships are invested in the same securities and in approximately the same percentages.
+However, particularly during the initial stages, money becomes available at varying times and varying levels of
+the market so there is more variation in results than is likely to be the case in later years. Over the years, I will
+be quite satisfied with a performance that is 10% per year better than the Averages, so in respect to these three
+partnerships, 1957 was a successful and probably better than average, year.
+Two partnerships were started during the middle of 1957 and their results for the balance of the year were
+roughly the same as the performance of the Averages which were down about 12% for the period since
+inception of the 1957 partnerships. Their portfolios are now starting to approximate those of the 1956
+partnerships and performance of the entire group should be much more comparable in the future.
+
+Interpretation of results
+
+To some extent our better than average performance in 1957 was due to the fact that it was a generally poor year
+for most stocks. Our performance, relatively, is likely to be better in a bear market than in a bull market so that
+deductions made from the above results should be tempered by the fact that it was the type of year when we
+should have done relatively well. In a year when the general market had a substantial advance I would be well
+satisfied to match the advance of the Averages.
+
+I can definitely say that our portfolio represents better value at the end of 1957 than it did at the end of 1956.
+This is due to both generally lower prices and the fact that we have had more time to acquire the more
+substantially undervalued securities which can only be acquired with patience. Earlier I mentioned our largest
+position which comprised 10% to 20% of the assets of the various partnerships. In time I plan to have this
+represent 20% of the assets of all partnerships but this cannot be hurried. Obviously during any acquisition
+period, our primary interest is to have the stock do nothing or decline rather than advance. Therefore, at any
+given time, a fair proportion of our portfolio may be in the sterile stage. This policy, while requiring patience,
+should maximize long term profits.
+
+I have tried to cover points which I felt might be of interest and disclose as much of our philosophy as may be
+imparted without talking of individual issues. If there are any questions concerning any phase of the operation, I
+would welcome hearing from you.

bpl/1957/bpl-1957-letter.txt

@@ -0,0 +1,92 @@
+1957 Letter
+
+WARREN E. BUFFETT
+
+5202 Underwood Ave. Omaha, Nebraska
+
+SECOND ANNUAL LETTER TO LIMITED PARTNERS
+
+The General Stock Market Picture in 1957
+
+In last year's letter to partners, I said the following:
+
+My view of the general market level is that it is priced above intrinsic value. This view relates
+to blue-chip securities. This view, if accurate, carries with it the possibility of a substantial
+decline in all stock prices, both undervalued and otherwise. In any event I think the probability
+is very slight that current market levels will be thought of as cheap five years from now. Even a
+full-scale bear market, however, should not hurt the market value of our work-outs substantially.
+
+If the general market were to return to an undervalued status our capital might be employed exclusively in
+general issues and perhaps some borrowed money would be used in this operation at that time. Conversely, if
+the market should go considerably higher our policy will be to reduce our general issues as profits present
+themselves and increase the work-out portfolio.
+
+All of the above is not intended to imply that market analysis is foremost in my mind. Primary attention is given
+at all times to the detection of substantially undervalued securities.
+The past year witnessed a moderate decline in stock prices. I stress the word "moderate" since casual reading of
+the press or conversing with those who have had only recent experience with stocks would tend to create an
+impression of a much greater decline. Actually, it appears to me that the decline in stock prices has been
+considerably less than the decline in corporate earning power under present business conditions. This means that
+the public is still very bullish on blue chip stocks and the general economic picture. I make no attempt to
+forecast either business or the stock market; the above is simply intended to dispel any notions that stocks have
+suffered any drastic decline or that the general market, is at a low level. I still consider the general market to be
+priced on the high side based on long term investment value.
+Our Activities in 1957
+The market decline has created greater opportunity among undervalued situations so that, generally, our
+portfolio is heavier in undervalued situations relative to work-outs than it was last year. Perhaps an explanation
+of the term "work-out" is in order. A work-out is an investment which is dependent on a specific corporate
+action for its profit rather than a general advance in the price of the stock as in the case of undervalued
+situations. Work-outs come about through: sales, mergers, liquidations, tenders, etc. In each case, the risk is that
+something will upset the applecart and cause the abandonment of the planned action, not that the economic
+picture will deteriorate and stocks decline generally. At the end of 1956, we had a ratio of about 70-30 between
+general issues and work-outs. Now it is about 85-15.
+During the past year we have taken positions in two situations which have reached a size where we may expect
+to take some part in corporate decisions. One of these positions accounts for between 10% and 20% of the
+portfolio of the various partnerships and the other accounts for about 5%. Both of these will probably take in the
+neighborhood of three to five years of work but they presently appear to have potential for a high average annual
+rate of return with a minimum of risk. While not in the classification of work-outs, they have very little
+1
+dependence on the general action of the stock market. Should the general market have a substantial rise, of
+course, I would expect this section of our portfolio to lag behind the action of the market.
+Results for 1957
+In 1957 the three partnerships which we formed in 1956 did substantially better than the general market. At the
+beginning of the year, the Dow-Jones Industrials stood at 499 and at the end of the year it was at 435 for a loss
+of 64 points. If one had owned the Averages, he would have received 22 points in dividends reducing the overall
+loss to 42 points or 8.470% for the year. This loss is roughly equivalent to what would have been achieved by
+investing in most investment funds and, to my knowledge, no investment fund invested in stocks showed a gain
+for the year.
+All three of the 1956 partnerships showed a gain during the year amounting to about 6.2%, 7.8% and 25% on
+yearend 1956 net worth. Naturally a question is created as to the vastly superior performance of the last
+partnership, particularly in the mind of the partners of the first two. This performance emphasizes the
+importance of luck in the short run, particularly in regard to when funds are received. The third partnership was
+started the latest in 1956 when the market was at a lower level and when several securities were particularly
+attractive. Because of the availability of funds, large positions were taken in these issues. Whereas the two
+partnerships formed earlier were already substantially invested so that they could only take relatively small
+positions in these issues.
+Basically, all partnerships are invested in the same securities and in approximately the same percentages.
+However, particularly during the initial stages, money becomes available at varying times and varying levels of
+the market so there is more variation in results than is likely to be the case in later years. Over the years, I will
+be quite satisfied with a performance that is 10% per year better than the Averages, so in respect to these three
+partnerships, 1957 was a successful and probably better than average, year.
+Two partnerships were started during the middle of 1957 and their results for the balance of the year were
+roughly the same as the performance of the Averages which were down about 12% for the period since
+inception of the 1957 partnerships. Their portfolios are now starting to approximate those of the 1956
+partnerships and performance of the entire group should be much more comparable in the future.
+Interpretation of results
+To some extent our better than average performance in 1957 was due to the fact that it was a generally poor year
+for most stocks. Our performance, relatively, is likely to be better in a bear market than in a bull market so that
+deductions made from the above results should be tempered by the fact that it was the type of year when we
+should have done relatively well. In a year when the general market had a substantial advance I would be well
+satisfied to match the advance of the Averages.
+I can definitely say that our portfolio represents better value at the end of 1957 than it did at the end of 1956.
+This is due to both generally lower prices and the fact that we have had more time to acquire the more
+substantially undervalued securities which can only be acquired with patience. Earlier I mentioned our largest
+position which comprised 10% to 20% of the assets of the various partnerships. In time I plan to have this
+represent 20% of the assets of all partnerships but this cannot be hurried. Obviously during any acquisition
+period, our primary interest is to have the stock do nothing or decline rather than advance. Therefore, at any
+given time, a fair proportion of our portfolio may be in the sterile stage. This policy, while requiring patience,
+should maximize long term profits.
+2
+I have tried to cover points which I felt might be of interest and disclose as much of our philosophy as may be
+imparted without talking of individual issues. If there are any questions concerning any phase of the operation, I
+would welcome hearing from you.

bpl/1957/inputs-bpl-1957.txt

@@ -0,0 +1,63 @@
+file 3-wavs/output-000.wav'
+file 3-wavs/output-001.wav'
+file 3-wavs/output-002.wav'
+file 3-wavs/output-003.wav'
+file 3-wavs/output-004.wav'
+file 3-wavs/output-005.wav'
+file 3-wavs/output-006.wav'
+file 3-wavs/output-007.wav'
+file 3-wavs/output-008.wav'
+file 3-wavs/output-009.wav'
+file 3-wavs/output-010.wav'
+file 3-wavs/output-011.wav'
+file 3-wavs/output-012.wav'
+file 3-wavs/output-013.wav'
+file 3-wavs/output-014.wav'
+file 3-wavs/output-015.wav'
+file 3-wavs/output-016.wav'
+file 3-wavs/output-017.wav'
+file 3-wavs/output-018.wav'
+file 3-wavs/output-019.wav'
+file 3-wavs/output-020.wav'
+file 3-wavs/output-021.wav'
+file 3-wavs/output-022.wav'
+file 3-wavs/output-023.wav'
+file 3-wavs/output-024.wav'
+file 3-wavs/output-025.wav'
+file 3-wavs/output-026.wav'
+file 3-wavs/output-027.wav'
+file 3-wavs/output-028.wav'
+file 3-wavs/output-029.wav'
+file 3-wavs/output-030.wav'
+file 3-wavs/output-031.wav'
+file 3-wavs/output-032.wav'
+file 3-wavs/output-033.wav'
+file 3-wavs/output-034.wav'
+file 3-wavs/output-035.wav'
+file 3-wavs/output-036.wav'
+file 3-wavs/output-037.wav'
+file 3-wavs/output-038.wav'
+file 3-wavs/output-039.wav'
+file 3-wavs/output-040.wav'
+file 3-wavs/output-041.wav'
+file 3-wavs/output-042.wav'
+file 3-wavs/output-043.wav'
+file 3-wavs/output-044.wav'
+file 3-wavs/output-045.wav'
+file 3-wavs/output-046.wav'
+file 3-wavs/output-047.wav'
+file 3-wavs/output-048.wav'
+file 3-wavs/output-049.wav'
+file 3-wavs/output-050.wav'
+file 3-wavs/output-051.wav'
+file 3-wavs/output-052.wav'
+file 3-wavs/output-053.wav'
+file 3-wavs/output-054.wav'
+file 3-wavs/output-055.wav'
+file 3-wavs/output-056.wav'
+file 3-wavs/output-057.wav'
+file 3-wavs/output-058.wav'
+file 3-wavs/output-059.wav'
+file 3-wavs/output-060.wav'
+file 3-wavs/output-061.wav'
+file 3-wavs/output-062.wav'

gen_sox_inputs.py

@@ -0,0 +1,26 @@
+import glob
+import sys
+import subprocess
+import os
+
+path = sys.argv[1]
+fn_stem = sys.argv[2]
+
+files = glob.glob(f"{path}/3-wavs/output-???.wav")
+files.sort()
+
+print(files)
+
+with open(f"./{path}/inputs-{fn_stem}.txt", "w") as f:
+    for file in files:
+        base_file = os.path.basename(file)
+        f.write(f"file 3-wavs/{base_file}'\n")
+    f.flush()
+
+subprocess.run([
+    "ffmpeg",
+    "-f", "concat",
+    "-safe", "0",
+    "-i", f"./{path}/inputs-{fn_stem}.txt",
+    "-c", "copy", f"./{path}/{fn_stem}.wav",
+    ])

speak.py

@@ -3,25 +3,47 @@ from urllib.parse import quote
 import subprocess
 import os
 import sys
+import re
 
 speaker_id="p230"
 
-letter_year = sys.argv[1]
+letter_series = sys.argv[1] or "buffett-partnership"
+letter_year = sys.argv[2] or "1956"
+real_run = (len(sys.argv) == 4) and bool(sys.argv[3]) or False
 
-fn = f"./annual-letters/bpl-{letter_year}.txt"
+fn = f"./{letter_series}/{letter_year}/1-in/bpl-{letter_year}-letter.txt"
+pre_fn = f"./{letter_series}/{letter_year}/2-preprocess/bpl-{letter_year}.txt"
 
-output_dir = f"./wavs/bpl-{letter_year}/"
+output_dir = f"./{letter_series}/{letter_year}/3-wavs"
 
 subprocess.run(['mkdir', '-p', output_dir])
 
+if (not real_run):
+    print("Dry run, check that sentences are readable.")
+
+def synth_sentence(sentence, i):
+    text_prompt = quote(sentence)
+    # If hosted on IPv6 use http://[::1]:5002
+    # If hosted on IPv4 use http://localhost:5002
+    query_string = f"http://\[::1\]:5002/api/tts?text={text_prompt}&style_wav=&language_id=&speaker_id={speaker_id}"
+    print(f"Query string {query_string}")
+    subprocess.run(['curl', query_string, '-o', f"./{output_dir}/output-{str(i).zfill(3)}.wav"])
+
 with open(fn) as f:
     lines = f.readlines()
     all_text = ''.join(lines)
-    sentences = all_text.split('.')
+    sentence_ends = list(re.finditer(r"(\w\w+([\.:;][\s\n])|([\s\n][\s\n]+))", all_text))
-    count = len(sentences)
     
-    for (i, sentence) in enumerate(sentences):
+    start = 0
-        text_prompt = quote(sentence)
+    count = len(sentence_ends)
-        query_string = f"http://localhost:5002/api/tts?text={text_prompt}&speaker_id={speaker_id}&style_wav=&language_id="
+    print(f"{count} sentences found.\n")
-        subprocess.run(['curl', query_string, '-o', f"./{output_dir}/output-{str(i).zfill(3)}.wav"])
+
+    for (i, sentence_end) in enumerate(sentence_ends):
+        sentence = all_text[start:sentence_end.end()].strip()
+        # Heal any newlines due to awkward breaks from copy-pasta
+        sentence = " ".join(map(lambda x: x.strip(), sentence.split("\n")))
+        print(f"Sentence {i}:\n\t\"{sentence}\"\n\t{sentence_end}")
+        start = sentence_end.end() # no +1 needed since end is already one past the real ending character
     
+        if (real_run):
+            synth_sentence(sentence, i)