Skip to content

Commit

Permalink
Add support for using confidences from utterances and words
Browse files Browse the repository at this point in the history
  • Loading branch information
@Alex-Cook4
Alex-Cook4 committed Jan 2, 2017
1 parent 31325be commit 2d32692
Show file tree
Hide file tree
Showing 8 changed files with 115 additions and 57 deletions.
6 changes: 6 additions & 0 deletions .project
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,13 @@
<projects>
</projects>
<buildSpec>
<buildCommand>
<name>org.eclipse.xtext.ui.shared.xtextBuilder</name>
<arguments>
</arguments>
</buildCommand>
</buildSpec>
<natures>
<nature>org.eclipse.xtext.ui.shared.xtextNature</nature>
</natures>
</projectDescription>
6 changes: 5 additions & 1 deletion applications/CallCenter/README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -8,4 +8,8 @@ the results via TCP.
This application can be modified to fit the size of your environment by providing compile-time
arguments for the number of hosts and the number of engines to run on each host.

Check out this video about how Verizon is using this solution: https://youtu.be/Zg-_BJt6jdc
Check out this video about how Verizon is using this solution: https://youtu.be/Zg-_BJt6jdc

Requirements:
1. atlas, atlas-devel, libcap-devel, libpcap-devel packages (use yum install)
2. com.ibm.streams.speech2text and com.ibm.streamsx.network toolkit.
21 changes: 15 additions & 6 deletions ...eech2text.speechprocessing/com.ibm.streamsx.speech2text.speechprocessing.outdata/Main.spl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,12 +30,14 @@ composite Main
Utterance = tuple<rstring callId,
int32 utteranceNumber,
float64 utteranceStartTime,
float64 utteranceEndTime,
uint32 captureSeconds,
rstring role,
float64 utteranceEndTime,
uint32 captureSeconds,
rstring role,
rstring utterance,
int32 speakerId,
rstring callCenter>;
rstring callCenter,
float64 utteranceConfidence,
list<float64> utteranceTokenConfidences>;


graph
Expand Down Expand Up @@ -87,7 +89,14 @@ composite Main
}
}


() as TestSink = FileSink(Import_Utterances){
param
file: "/home/streamsadmin/data/test.dat";
format : csv ;
closeMode : time;
timePerFile : 86000.0f;
flush : 10u;
}

() as TcpSink_Import_UtterancesStream = TCPSink(Import_Utterances)
{
Expand All @@ -103,7 +112,7 @@ composite Main
}

config
hostPool : P1=createPool({size=1u, tags=["setup"]},Sys.Shared);
hostPool : P1=createPool({size=1u, tags=["watson5"]},Sys.Shared);
placement : host(P1);
restartable : false;

Expand Down
19 changes: 14 additions & 5 deletions ...ch2text.speechprocessing/com.ibm.streamsx.speech2text.speechprocessing.outdata/Main.splmm
View file
Original file line number Diff line number Diff line change
Expand Up @@ -39,12 +39,14 @@ composite Main
Utterance = tuple<rstring callId,
int32 utteranceNumber,
float64 utteranceStartTime,
float64 utteranceEndTime,
uint32 captureSeconds,
rstring role,
float64 utteranceEndTime,
uint32 captureSeconds,
rstring role,
rstring utterance,
int32 speakerId,
rstring callCenter>;
rstring callCenter,
float64 utteranceConfidence,
list<float64> utteranceTokenConfidences>;


graph
Expand Down Expand Up @@ -96,7 +98,14 @@ composite Main
}
}


() as TestSink = FileSink(Import_Utterances){
param
file: "/home/streamsadmin/data/test.dat";
format : csv ;
closeMode : time;
timePerFile : 86000.0f;
flush : 10u;
}

() as TcpSink_Import_UtterancesStream = TCPSink(Import_Utterances)
{
Expand Down
2 changes: 1 addition & 1 deletion ...speech2text.speechprocessing/com.ibm.streamsx.speech2text.speechprocessing.setup/Main.spl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -583,7 +583,7 @@ composite Main


config
hostPool : P1=createPool({size=1u, tags=["setup"]},Sys.Shared);
hostPool : P1=createPool({size=1u, tags=["watson5"]},Sys.Shared);
placement : host(P1) , hostColocation("SameHost");


Expand Down
73 changes: 44 additions & 29 deletions ...msx.speech2text.speechprocessing/com.ibm.streamsx.speech2text.speechprocessing/Watson.spl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -21,10 +21,23 @@ use com.ibm.streamsx.network.parse::* ;
use com.ibm.streams.speech2text.watson::*;


type bestUtterance = tuple<rstring id, rstring channel_id, float64 utteranceStartTime, float64 utteranceEndTime, uint32 captureSeconds, int32 utteranceNumber, rstring utterance> ;

type Utterance = tuple<rstring callId, int32 utteranceNumber, float64 utteranceStartTime, float64 utteranceEndTime, uint32 captureSeconds, rstring role, rstring utterance, int32 speakerId, rstring callCenter>;
type Utterance_File = tuple<rstring ipaddress, rstring startingTime, rstring callCenter, int32 utteranceNumber, float64 utteranceStartTime, float64 utteranceEndTime, uint32 captureSeconds, rstring role, rstring utterance, int32 speakerId>;
type bestUtterance = tuple<rstring id, rstring channel_id
, float64 utteranceStartTime, float64 utteranceEndTime
, uint32 captureSeconds, int32 utteranceNumber
, rstring utterance, float64 utteranceConfidence
, list<float64> utteranceTokenConfidences> ;

type Utterance = tuple<rstring callId, int32 utteranceNumber
, float64 utteranceStartTime, float64 utteranceEndTime
, uint32 captureSeconds, rstring role
, rstring utterance, int32 speakerId
, rstring callCenter, float64 utteranceConfidence
, list<float64> utteranceTokenConfidences>;
type Utterance_File = tuple<rstring ipaddress, rstring startingTime
, rstring callCenter, int32 utteranceNumber
, float64 utteranceStartTime, float64 utteranceEndTime
, uint32 captureSeconds, rstring role
, rstring utterance, int32 speakerId>;


composite Watson
Expand Down Expand Up @@ -79,22 +92,22 @@ composite Watson



stream<rstring id, rstring channel_id, uint32 ssrc, uint32 ts, uint32 captureSeconds, blob speech, int32 samples, int32 jobId, uint32 hostId> Imported_Utterances1 = Import() {
stream<rstring id, rstring channel_id, uint32 ssrc, uint32 ts, uint32 captureSeconds, blob speech, int32 samples, int32 jobId, uint32 hostId> Imported_Utterances5 = Import() {
param
applicationName : "com.ibm.streamsx.speech2text.speechprocessing.setup::Main";
streamId : "Import_Utterances1" ;
streamId : "Import_Utterances5" ;
}


stream<rstring id, rstring channel_id, uint32 ssrc, uint32 ts, uint32 captureSeconds, blob speech, int32 samples, int32 jobId, uint32 hostId> Imported2_Utterances1 = Functor(Imported_Utterances1) {
stream<rstring id, rstring channel_id, uint32 ssrc, uint32 ts, uint32 captureSeconds, blob speech, int32 samples, int32 jobId, uint32 hostId> Imported2_Utterances5 = Functor(Imported_Utterances5) {
}


@parallel(width = 50 , partitionBy = [ { port = Imported2_Utterances1, attributes = [jobId] } ])
@parallel(width = 3 , partitionBy = [ { port = Imported2_Utterances5, attributes = [jobId] } ])
(
stream<bestUtterance> Utterances1;
stream<rstring log> Diagnostics1
) = SpeechToText(Imported2_Utterances1)
stream<bestUtterance> Utterances5;
stream<rstring log> Diagnostics5
) = SpeechToText(Imported2_Utterances5)
{
param
paramsetFile : $paramsetFile;
Expand All @@ -103,7 +116,7 @@ composite Watson

/*

stream<bestUtterance> Filtered_Utterances1 = Filter(Utterances1)
stream<bestUtterance> Filtered_Utterances5 = Filter(Utterances5)
{
param filter : ( channel_id == "0" | channel_id == "1" ) ;

Expand All @@ -113,19 +126,19 @@ composite Watson


(
stream<Utterance> Utterance_Data1 ;
stream<Utterance_File> Utterance_File1 )
= Custom(Utterances1)
stream<Utterance> Utterance_Data5 ;
stream<Utterance_File> Utterance_File5 )
= Custom(Utterances5)
{
logic
onTuple Utterances1 :
onTuple Utterances5 :
{
list<rstring> tokens = tokenize(id,"_",true);

// mutable rstring tempUtterance = utterance;
submit( {callId = $callCenter + "_" + id, utteranceNumber = utteranceNumber, utteranceStartTime = utteranceStartTime, utteranceEndTime = utteranceEndTime, captureSeconds = captureSeconds, role = "AGENT", utterance = utterance, speakerId = (int32)channel_id, callCenter = $callCenter } , Utterance_Data1 );
submit( {callId = $callCenter + "_" + id, utteranceNumber = utteranceNumber, utteranceStartTime = utteranceStartTime, utteranceEndTime = utteranceEndTime, captureSeconds = captureSeconds, role = "AGENT", utterance = utterance, speakerId = (int32)channel_id, callCenter = $callCenter, utteranceConfidence = utteranceConfidence, utteranceTokenConfidences = utteranceTokenConfidences } , Utterance_Data5 );
submit( {ipaddress = tokens[0], startingTime = tokens[1], callCenter = $callCenter, utteranceNumber = utteranceNumber,
utteranceStartTime = utteranceStartTime, utteranceEndTime = utteranceEndTime, captureSeconds = captureSeconds, role = "AGENT", utterance = utterance, speakerId = (int32)channel_id } , Utterance_File1 );
utteranceStartTime = utteranceStartTime, utteranceEndTime = utteranceEndTime, captureSeconds = captureSeconds, role = "AGENT", utterance = utterance, speakerId = (int32)channel_id } , Utterance_File5 );


}
Expand All @@ -134,37 +147,37 @@ composite Watson

// jmsjms : This export needs to eventually go away and be replaced with a messging system

() as ExportFileOutput1 = Export(Utterance_Data1)
() as ExportFileOutput5 = Export(Utterance_Data5)
{
param
streamId : "Utterance_Data1";
streamId : "Utterance_Data5";
}


() as FileOutput1 = FileSink(Utterance_File1 )
() as FileOutput5 = FileSink(Utterance_File5 )
{
param
file: $utterances + "outUtterances1_{localtime:%m%d%y}_{id}.csv";
file: $utterances + "outUtterances5_{localtime:%m%d%y}_{id}.csv";
format : csv ;
closeMode : time;
timePerFile : 86000.0f;
flush : 1u;
flush : 10u;

}

() as FileDiagnostics1 = FileSink(Diagnostics1 )
() as FileDiagnostics5 = FileSink(Diagnostics5 )
{
param
file: $diagnostics + "Diagnostics1_{localtime:%m%d%y}_{id}.csv";
file: $diagnostics + "Diagnostics5_{localtime:%m%d%y}_{id}.csv";
format : csv ;
closeMode : time;
timePerFile : 86000.0f;
flush : 1u;
flush : 10u;
}


config
hostPool : P1=createPool({size=1u, tags=["setup"]},Sys.Shared);
hostPool : P1=createPool({size=1u, tags=["watson5"]},Sys.Shared);
placement : host(P1);
}

Expand All @@ -187,9 +200,11 @@ composite SpeechToText(input inp ; output Utterances, Debug)
output
Utterances :
utteranceStartTime = getUtteranceStartTime(),
utteranceEndTime = getUtteranceEndTime(),
utteranceEndTime = getUtteranceEndTime(),
utteranceNumber = getUtteranceNumber(),
utterance = getUtteranceText();
utterance = getUtteranceText(),
utteranceConfidence = getUtteranceConfidence(),
utteranceTokenConfidences = getUtteranceWordConfidences();
Debug :
log = getDiagnosticsMessage();
config
Expand Down
33 changes: 24 additions & 9 deletions ...x.speech2text.speechprocessing/com.ibm.streamsx.speech2text.speechprocessing/Watson.splmm
View file
Original file line number Diff line number Diff line change
Expand Up @@ -31,10 +31,23 @@ use com.ibm.streamsx.network.parse::* ;
use com.ibm.streams.speech2text.watson::*;


type bestUtterance = tuple<rstring id, rstring channel_id, float64 utteranceStartTime, float64 utteranceEndTime, uint32 captureSeconds, int32 utteranceNumber, rstring utterance> ;

type Utterance = tuple<rstring callId, int32 utteranceNumber, float64 utteranceStartTime, float64 utteranceEndTime, uint32 captureSeconds, rstring role, rstring utterance, int32 speakerId, rstring callCenter>;
type Utterance_File = tuple<rstring ipaddress, rstring startingTime, rstring callCenter, int32 utteranceNumber, float64 utteranceStartTime, float64 utteranceEndTime, uint32 captureSeconds, rstring role, rstring utterance, int32 speakerId>;
type bestUtterance = tuple<rstring id, rstring channel_id
, float64 utteranceStartTime, float64 utteranceEndTime
, uint32 captureSeconds, int32 utteranceNumber
, rstring utterance, float64 utteranceConfidence
, list<float64> utteranceTokenConfidences> ;

type Utterance = tuple<rstring callId, int32 utteranceNumber
, float64 utteranceStartTime, float64 utteranceEndTime
, uint32 captureSeconds, rstring role
, rstring utterance, int32 speakerId
, rstring callCenter, float64 utteranceConfidence
, list<float64> utteranceTokenConfidences>;
type Utterance_File = tuple<rstring ipaddress, rstring startingTime
, rstring callCenter, int32 utteranceNumber
, float64 utteranceStartTime, float64 utteranceEndTime
, uint32 captureSeconds, rstring role
, rstring utterance, int32 speakerId>;


composite Watson
Expand Down Expand Up @@ -136,7 +149,7 @@ composite Watson
list<rstring> tokens = tokenize(id,"_",true);

// mutable rstring tempUtterance = utterance;
submit( {callId = $callCenter + "_" + id, utteranceNumber = utteranceNumber, utteranceStartTime = utteranceStartTime, utteranceEndTime = utteranceEndTime, captureSeconds = captureSeconds, role = "AGENT", utterance = utterance, speakerId = (int32)channel_id, callCenter = $callCenter } , Utterance_Data<%=$jobNumber%> );
submit( {callId = $callCenter + "_" + id, utteranceNumber = utteranceNumber, utteranceStartTime = utteranceStartTime, utteranceEndTime = utteranceEndTime, captureSeconds = captureSeconds, role = "AGENT", utterance = utterance, speakerId = (int32)channel_id, callCenter = $callCenter, utteranceConfidence = utteranceConfidence, utteranceTokenConfidences = utteranceTokenConfidences } , Utterance_Data<%=$jobNumber%> );
submit( {ipaddress = tokens[0], startingTime = tokens[1], callCenter = $callCenter, utteranceNumber = utteranceNumber,
utteranceStartTime = utteranceStartTime, utteranceEndTime = utteranceEndTime, captureSeconds = captureSeconds, role = "AGENT", utterance = utterance, speakerId = (int32)channel_id } , Utterance_File<%=$jobNumber%> );

Expand All @@ -161,7 +174,7 @@ composite Watson
format : csv ;
closeMode : time;
timePerFile : 86000.0f;
flush : 1u;
flush : 10u;

}

Expand All @@ -172,7 +185,7 @@ composite Watson
format : csv ;
closeMode : time;
timePerFile : 86000.0f;
flush : 1u;
flush : 10u;
}


Expand Down Expand Up @@ -200,9 +213,11 @@ composite SpeechToText(input inp ; output Utterances, Debug)
output
Utterances :
utteranceStartTime = getUtteranceStartTime(),
utteranceEndTime = getUtteranceEndTime(),
utteranceEndTime = getUtteranceEndTime(),
utteranceNumber = getUtteranceNumber(),
utterance = getUtteranceText();
utterance = getUtteranceText(),
utteranceConfidence = getUtteranceConfidence(),
utteranceTokenConfidences = getUtteranceWordConfidences();
Debug :
log = getDiagnosticsMessage();
config
Expand Down
Loading

0 comments on commit 2d32692

Please sign in to comment.