কত কথা যে বলতে চাই........

python integer to string

Firoj Alam — Mon, 11 Apr 2011 01:36:04 +0000

str() function does not always work. We need something else to solve the “TypeError: ‘str’ object is not callable python” problem while using str(). Put “%s”%i
i = 0
i = i + 1
str = str+” >$pathfst/S”+”%s”%i+”.fst”
print str

sed useful command

Firoj Alam — Sat, 02 Apr 2011 09:31:33 +0000

To select specific number of lines this command is really helpful.
sed -n -e 1,1091p -e 1093,1205p -e 1207,1527p -e 1527,1539p -e 1541,1555p -e 1557,1566p -e 1559,1571p -e 1574,1875p

unit selection build clunits

Firoj Alam — Thu, 07 Jan 2010 06:45:28 +0000

unit selection build clunits
festvox-clunits

Load all the utterances identified in the database. festival/utts/*.utt
Find the same stem item and collect them into lists of that type. find_same_types utterances params
- set the unittypes.
Set the unit types in name_units: name each unit with a unique id and number the occurances of each type. It is based on festival/clunits/all.desc file. eg. unit= “k_V” id=5703. Also count the total units.
Dump features: dump_feats unittypes utterances clunits_params
- dumping the features for clustering
- Do multiple passes over utterances for each unittype and dump the desired features. This would be easier if utterances would not require feature functions.
- Dumping features for all items of type. eg; dumping feature for unit = k_V in festival/feats/*.feat
Load coeffs: utts_load_coeffs: loading the acoustic coefficients for each utterances.
Build disttabs an dcluster: Build each table individually and cluster them. Remooving the table before moving on to the next step.
1. disttabs_and_cluster unittypes clunits_params
  1. 1. unittypes
      
      clunits params: these params typically appears in festival/tree/INTS_LANG_VOX.tree.
    - Requires
    - It
      generates the distance table (disttab) files from the feature (*.feats)
      file for each unit type. The distance table file is going to be too
      large and there is a mechanism to delete the file after generating the
      tree for the unit.
    - It uses the wagon (available in speech_tools) to find the best cluster.
    - Build the tree using wagon for each unit type
Collect tree: collect_trees unittypes params
- collect the trees into one files as an assoc list.
Save catalogue file: save_catalogue utterances clunits_params

আমাদের স্বপ্ন

Firoj Alam — Sat, 02 Jan 2010 03:28:25 +0000

স্বপ্ন শপথ, প্রথম আলো- ছুটির দিনে, ২ জানুয়ারী ২০১০

http://www.prothom-alo.com/detail/date/2010-01-02/news/31920

Versioning system i like to follow….

Firoj Alam — Sat, 19 Dec 2009 02:19:31 +0000

I am adhere to the following versioning system:

MAJOR.MINOR.PATCH
1.1.9

Major: Any kind of change

remove or change constants
remove (deprecated) functions
fold together macro-ized function replacements

Minor:

Minor releases can introduce:

new functions,
new symbolic and enumerated constants,
deprecate existing functions.

Patch/update:

Can only change function implementations
These releases are pure bug fix releases
Changes to the API, to the signatures of public functions, or to the interpretation of function parameters is not allowed.

Ref:

http://unicode.org/versions/

http://apr.apache.org/versioning.html

Phonotactic constrains – diphone

Firoj Alam — Mon, 14 Dec 2009 15:33:35 +0000

This could perhaps a thought to minimize the diphone set. On the other hand i like to say defining the phonotactic constrain in Bangla. Some possible solutions could be the following:
1. Consonants + monophthong vowel + most frequent diphthong (ঐ and ঔ) -> 30 + 14 + 2 = 46. So 46*46 = 2116 diphones coverage corpus can provide a optimal solution. A backoff procedure can be use for rest of the diphthongs.
2. For a small context, only consonant and neutral vowel can provide a good solution.

The best solution needs significant amount of manual effort by linguistic expert to define phonotactic constrain.

Completely lost

Firoj Alam — Sat, 12 Dec 2009 13:38:39 +0000

I don’t know, what happen to me. I’m not getting anything to do, though i have a lot of work, a “to do” list in front of me. Probably, I’m suffering from a stress. Sometime i do feel the closest people surround me are no longer closer. Feeling myself as a misanthropic person. I don’t know, why. However, I’m not a cynical at all, I’m in a category of ambivert, philanthropic, believe in god, like to define luck as a best attempt, making lot of mistakes when write. Anyway, trying to improve.

Bangla DANDA and DOUBLE DANDA

Firoj Alam — Sun, 28 Jun 2009 02:20:07 +0000

DANDA is e0 a5 a4 and DOUBLE DANDA is e0 a5 a5

A little hack:

for (j = 0; j < i; )
{
// DANDA is e0 a5 a4 and DOUBLE DANDA is e0 a5 a5

if ((unsigned char)word[j] == 0xe0 && j < i-2)
{
if ((unsigned char)word[j+1] == 0xa5 &&
((unsigned char)word[j+2] == 0xa4 ||
(unsigned char)word[j+2] == 0xa5))
{
word[j] = word[j+1] = word[j+2] = ‘;’;
j += 3;
}
else
{
j++;
}
}
else
{
j++;
}
}

// using decimal val: 1st byte 224, 2nd byte 166, 167 and 3rd byte 127, 191 which identifies the value is Bangla unicode.

1st byte:
decimal: 224
octal: oct340
hex: e0

2nd byte:
decimal: 166 and 167
octal: oct246 and oct247
hex: a6 and a7

3rd byte:
decimal: 128 to 191
octal: oct200 to oct277
hex: 80 to b9

for (j = 0; j < i; )
{
// DANDA is e0 a5 a4
if (ord(word[j]) == 224 && j 127 && ord(word[j+2])<192))
{
return true;
break;
}
else
{
j++;
}
}
else
{
j++;
}
}

গ্রেডিং হিসাব

Firoj Alam — Sat, 27 Jun 2009 14:05:15 +0000

SSC and HSC

class interval	Letter Grade	Grade point
80-100	A+	5
70-79	A	4
60-69	A-	3.5
50-59	B	3
40-49	C	2
33-39	D	1
00-32	F	0

……Out of 5….

=IF(F2>79,”A+”,IF(AND(F2>69,F2<80),”A”, IF(AND(F2>59,F2<70),”A-”, IF(AND(F2>49,F2<60),”B”, IF(AND(F2>39,F2<50),”C”, IF(AND(F2>32,F2<40),”D”, “F”))))))

..Out of 4..

=IF(F2>79,”5″,IF(AND(F2>69,F2<80),”4″, IF(AND(F2>59,F2<70),”3.5″, IF(AND(F2>49,F2<60),”3″, IF(AND(F2>39,F2<50),”2″, IF(AND(F2>32,F2<40),”1″, “0″))))))

class interval	Letter Grade	Grade point
80-100	A+	5
70-79	A	4
60-69	A-	3.5
50-59	B	3
40-49	C	2
33-39	D	1
00-32	F	0

Conference Poster Standard Guideline

Firoj Alam — Sun, 18 Jan 2009 08:51:23 +0000

1. Poster should be readable from at least 5 feet. 2. 2.4m wide x 1.2m tall (8 x 4 ft) board will be provided for poster presentation. 3. I can use a single slide, enlarged and printed on a large-format printer. Otherwise i can use multiple slide and posted in sequence. 4. The title of my paper should appear in block letters which are 2.5 to 5 cm high (1-2 in, 110-220 point). The remaining text should be at least 1 to 2 cm high (0.4-0.8 in, 44-88 pt) to be easily legible to viewers who may stand as far as 2 meters away. Each graph or chart should be at least 27 x 21 cm (11 x 8.5 in). 4. The size of the poster should be A0 which is (1189×841 mm) corresponds to (46.81 inches height and 33.41 inches width) that means: approximately 3 A4 size paper in width and 6 A4 size in height.

Content:

/* Style Definitions */
table.MsoNormalTable
{mso-style-name:”Table Normal”;
mso-tstyle-rowband-size:0;
mso-tstyle-colband-size:0;
mso-style-noshow:yes;
mso-style-priority:99;
mso-style-qformat:yes;
mso-style-parent:”";
mso-padding-alt:0in 5.4pt 0in 5.4pt;
mso-para-margin-top:0in;
mso-para-margin-right:0in;
mso-para-margin-bottom:10.0pt;
mso-para-margin-left:0in;
line-height:115%;
mso-pagination:widow-orphan;
font-size:11.0pt;
font-family:”Calibri”,”sans-serif”;
mso-ascii-font-family:Calibri;
mso-ascii-theme-font:minor-latin;
mso-hansi-font-family:Calibri;
mso-hansi-theme-font:minor-latin;
mso-bidi-language:AR-SA;}

Possible sections
- Author/Title/Affiliation
- Poster Number
- Abstract
- Purpose or Hypothesis
- Background
- Data Collection and Analysis
- Objectives
- Data Sources/Study Setting
- Study Design
- Participants/Subjects
- Intervention
- Principal Findings
- Main Outcome Measures
- Results
- Discussion
- Implications
- Conclusions
- Funding Source

Ref:http://www.rfic2008.org/rfic2008/ifguide.html