[AsteriskBrasil] Usando o speech-recog.agi
Nuno Cunha
nuno.cunha em engdb.com.br
Quinta Fevereiro 1 15:26:44 BRST 2018
Ola José,
Mesmo comportamento!
Não vejo algum erro na chamada do google recognizer no CLI... apenas meu
AGI pegando o valor da variável "confidence" (que deve ser = 0...) e o
playback de endereço não compreendido.
Forcei uma chave errada e vi que não há informação sobre isto no debug
do console do Asterisk...
Coloquei aqui o que saiu no CLI. Se tiver um tempo para dar uma olhada e
ver se há algo estranho ai, agradeço demais.
Obrigado,
<PJSIP/6001-00000000>AGI Rx << EXEC "Playback" "PerguntarEnderecoOrigem"
-- AGI Script Executing Application: (Playback) Options:
(PerguntarEnderecoOrigem)
-- <PJSIP/6001-00000000> Playing 'PerguntarEnderecoOrigem.gsm'
(language 'en')
<PJSIP/6001-00000000>AGI Tx >> 200 result=0
<PJSIP/6001-00000000>AGI Rx << EXEC "Agi" "speech-recog.agi,pt-BR" --
AGI Script Executing Application: (Agi) Options: (speech-recog.agi,pt-BR)
-- Launched AGI Script /var/lib/asterisk/agi-bin/speech-recog.agi
<PJSIP/6001-00000000>AGI Tx >> agi_request: speech-recog.agi
<PJSIP/6001-00000000>AGI Tx >> agi_channel: PJSIP/6001-00000000
<PJSIP/6001-00000000>AGI Tx >> agi_language: en
<PJSIP/6001-00000000>AGI Tx >> agi_type: PJSIP
<PJSIP/6001-00000000>AGI Tx >> agi_uniqueid: 1517504895.0
<PJSIP/6001-00000000>AGI Tx >> agi_version: 13.19.0
<PJSIP/6001-00000000>AGI Tx >> agi_callerid: 6001
<PJSIP/6001-00000000>AGI Tx >> agi_calleridname: Nuno
<PJSIP/6001-00000000>AGI Tx >> agi_callingpres: 0
<PJSIP/6001-00000000>AGI Tx >> agi_callingani2: 0
<PJSIP/6001-00000000>AGI Tx >> agi_callington: 0
<PJSIP/6001-00000000>AGI Tx >> agi_callingtns: 0
<PJSIP/6001-00000000>AGI Tx >> agi_dnid: 100
<PJSIP/6001-00000000>AGI Tx >> agi_rdnis: unknown
<PJSIP/6001-00000000>AGI Tx >> agi_context: SolicitarEnderecosOrigemDestino
<PJSIP/6001-00000000>AGI Tx >> agi_extension: s
<PJSIP/6001-00000000>AGI Tx >> agi_priority: 2
<PJSIP/6001-00000000>AGI Tx >> agi_enhanced: 0.0
<PJSIP/6001-00000000>AGI Tx >> agi_accountcode:
<PJSIP/6001-00000000>AGI Tx >> agi_threadid: 139981899122432
<PJSIP/6001-00000000>AGI Tx >> agi_arg_1: pt-BR
<PJSIP/6001-00000000>AGI Tx >>
-- <PJSIP/6001-00000000>AGI Script speech-recog.agi completed,
returning 0
<PJSIP/6001-00000000>AGI Tx >> 200 result=0
<PJSIP/6001-00000000>AGI Rx << GET VARIABLE "confidence"
<PJSIP/6001-00000000>AGI Tx >> 200 result=0
<PJSIP/6001-00000000>AGI Rx << EXEC "Playback"
"AvisoDesculpeNaoCompreendiEndereco"
-- AGI Script Executing Application: (Playback) Options:
(AvisoDesculpeNaoCompreendiEndereco)
-- <PJSIP/6001-00000000> Playing
'AvisoDesculpeNaoCompreendiEndereco.gsm' (language 'en')
<PJSIP/6001-00000000>AGI Tx >> 200 result=0
----
Nuno Correia dos Santos Cunha
Analista de sistemas
Automação & Controle - Serviços Logann - T&T
Office:
Skype:
(31) 3211-7396
nuno.cunha.engdb
<http://www.engdb.com.br/> <https://www.facebook.com/engdb/>
<https://www.linkedin.com/company-beta/27862?pathWildcard=27862>
<https://www.youtube.com/channel/UCEUvHtjrY3-kD1tFryzCJyA> Avenida
Getúlio Vargas, nº 1.300, 9º e 10º andares | Belo Horizonte – MG | CEP
30112-021
2018-02-01 14:47 GMT-02:00 Jose <jasanchez113 em gmail.com>:
> Essa funciona so coloque sua chave:
>
>
> #!/usr/bin/env perl
>
>
> use warnings;
> use strict;
> use URI::Escape;
> use File::Copy qw(move);
> use File::Temp qw(tempfile);
> use LWP::UserAgent;
> use JSON;
> use Encode qw(encode);
>
> $| = 1;
>
> # ----------------------------- #
> # User defined parameters: #
> # ----------------------------- #
> # Speech API key #
> my $key = sua chave do google";
>
> # Default language #
> my $language = "pt-BR";
>
> # Default max silence timeout #
> my $timeout = 2;
>
> # Absolute Recording timeout #
> my $abs_timeout = -1;
>
> # Default interrupt key #
> my $intkey = "#";
>
> # Input audio sample rate #
> # Leave blank to auto-detect #
> my $samplerate = "";
>
> # Profanity filter #
> my $pro_filter = 0;
>
> # Use speex #
> my $use_speex = 0;
>
> # Use SSL #
> my $use_ssl = 1;
>
> # Verbose debugging messages #
> my $debug = 0;
>
> # ----------------------------- #
>
> my %AGI;
> my $ua;
> my $fh;
> my $tmpname;
> my $format;
> my @result;
> my $name;
> my $audio;
> my $uaresponse;
> my %response;
> my $endian;
> my $url;
> my $silence;
> my $filetype;
> my $flac;
> my $speex;
> my $results = 1;
> my $grammar = "builtin:dictation"; #"builtin:search";
> my $beep = "BEEP";
> my $comp_level = -8;
> my $ua_timeout = 10;
> my $tmpdir = "/tmp";
> my $host = "www.google.com/speech-api/v2/recognize";
>
> # Store AGI input #
> ($AGI{arg_1}, $AGI{arg_2}, $AGI{arg_3}, $AGI{arg_4}) = @ARGV;
> while (<STDIN>) {
> chomp;
> last if (!length);
> $AGI{$1} = $2 if (/^agi_(\w+)\:\s+(.*)$/);
> }
>
> $name = " -- $AGI{request}:";
>
> # Reset variables. #
> %response = (
> utterance => -1,
> confidence => -1,
> );
> warn "$name Clearing channel variables.\n" if ($debug);
> foreach (keys %response) {
> print "SET VARIABLE \"$_\" \"$response{$_}\"\n";
> checkresponse();
> }
>
> # Abort if key is missing or required programs not found. #
> if (!$key) {
> print "VERBOSE \"API key is missing. Aborting.\" 3\n";
> checkresponse();
> die "$name API key is missing. Aborting.\n";
> }
> if ($use_speex) {
> $speex = `/usr/bin/which speexenc`;
> die "$name speexenc is missing. Aborting.\n" if (!$speex);
> chomp($speex);
> warn "$name Found speexenc in: $speex\n" if ($debug);
> } else {
> $flac = `/usr/bin/which flac`;
> die "$name flac is missing. Aborting.\n" if (!$flac);
> chomp($flac);
> warn "$name Found flac in: $flac\n" if ($debug);
> }
>
> # Setting language, timeout, interrupt keys and BEEP indication #
> if (length($AGI{arg_1})) {
> $language = $AGI{arg_1} if ($AGI{arg_1} =~ /^[a-z]{2}(-[a-zA-Z]{2,6})?$/);
> }
>
> if (length($AGI{arg_2})) {
> if ($AGI{arg_2} == -1) {
> $silence = "";
> } elsif ($AGI{arg_2} =~ /^\d+$/) {
> $silence = "s=$AGI{arg_2}";
> } else {
> $silence = "s=$timeout";
> }
> } else {
> $silence = "s=$timeout";
> }
>
> if (length($AGI{arg_3})) {
> $intkey = "0123456789#*" if ($AGI{arg_3} eq "any");
> $intkey = $AGI{arg_3} if ($AGI{arg_3} =~ /^[0-9*#]+$/);
> }
>
> if (length($AGI{arg_4})) {
> $beep = "" if ($AGI{arg_4} eq "NOBEEP");
> }
>
> # Answer channel if not already answered #
> warn "$name Checking channel status.\n" if ($debug);
> print "CHANNEL STATUS\n";
> @result = checkresponse();
> if ($result[0] == 4) {
> warn "$name Answering channel.\n" if ($debug);
> print "ANSWER\n";
> @result = checkresponse();
> if ($result[0] != 0) {
> die "$name Failed to answer channel.\n";
> }
> }
>
> # Setting recording file format according to sample rate. #
> if (!$samplerate) { ($format, $samplerate) = detect_format(); }
> elsif ($samplerate == 12000) { $format = "sln12"; }
> elsif ($samplerate == 16000) { $format = "sln16"; }
> elsif ($samplerate == 32000) { $format = "sln32"; }
> elsif ($samplerate == 44100) { $format = "sln44"; }
> elsif ($samplerate == 48000) { $format = "sln48"; }
> else { ($format, $samplerate) = ("sln", 8000); }
>
> # Initialise User angent #
> if ($use_ssl) {
> $url = "https://" . $host;
> $ua = LWP::UserAgent->new(ssl_opts => {verify_hostname => 1});
> } else {
> $url = "http://" . $host;
> $ua = LWP::UserAgent->new;
> }
> $language = uri_escape($language);
> $grammar = uri_escape($grammar);
> $url .= "?key=$key&lang=$language&pfilter=$pro_filter&lm=$
> grammar&maxresults=$results";
> $ua->agent("Asterisk AGI speeech recognition script");
> $ua->env_proxy;
> $ua->timeout($ua_timeout);
>
> # Hnadle interrupts #
> $SIG{'INT'} = \&int_handler;
> $SIG{'HUP'} = \&int_handler;
>
> # Record file #
> ($fh, $tmpname) = tempfile("stt_XXXXXX", DIR => $tmpdir, UNLINK => 1);
> print "RECORD FILE $tmpname $format \"$intkey\" \"$abs_timeout\" $beep
> \"$silence\"\n";
> @result = checkresponse();
> die "$name Failed to record file, aborting...\n" if ($result[0] == -1);
>
> if ($debug) {
> warn "$name Recording Format: $format, Rate: $samplerate Hz, ",
> "Encoding format: ", ($use_speex) ? "speex" : "flac", "\n",
> "$name Languge: $language, SSL: ", ($use_ssl) ? "yes, " : "no, ",
> "$silence, Interrupt keys: $intkey\n";
> }
>
> # Encode sound data #
> if ($use_speex) {
> $filetype = "x-speex-with-header-byte";
> $endian = (unpack("h*", pack("s", 1)) =~ /01/) ? "--be" : "--le";
> # Encode file to speex. #
> system($speex, "--vbr", "--rate", $samplerate, "--headerbyte", "--quiet",
> $endian,
> "$tmpname.$format", "$tmpname.spx") == 0 or die "$name $speex failed:
> $?\n";
> open($fh, "<", "$tmpname.spx") or die "Can't read file: $!";
> } else {
> $filetype = "x-flac";
> $endian = (unpack("h*", pack("s", 1)) =~ /01/) ? "big" : "little";
> # Encode file to flac. #
> system($flac, $comp_level, "--totally-silent", "--channels=1",
> "--endian=$endian",
> "--sign=signed", "--bps=16", "--force-raw-format",
> "--sample-rate=$samplerate",
> "$tmpname.$format") == 0 or die "$name $flac failed: $?\n";
> open($fh, "<", "$tmpname.flac") or die "Can't read file: $!";
> }
>
> $audio = do { local $/; <$fh> };
> close($fh);
>
> # Send adio data for analysis #
> $uaresponse = $ua->post(
> "$url",
> Content_Type => "audio/$filetype; rate=$samplerate",
> Content => "$audio",
> );
> if (!$uaresponse->is_success) {
> print "VERBOSE \"Unable to get recognition data.\" 3\n";
> checkresponse();
> die "$name Unable to get recognition data.\n";
> }
>
> foreach (split(/\n/,$uaresponse->content)) {
> my $jdata = decode_json($_);
> for ( $jdata->{result}[0]->{alternative}[0] ) {
> $response{utterance} = encode('utf8', $_->{transcript});
> $response{confidence} = $_->{confidence};
> }
> }
> warn "$name The response was:\n", $uaresponse->content if ($debug);
>
> foreach (keys %response) {
> warn "$name Setting variable: $_ = $response{$_}\n" if ($debug);
> print "SET VARIABLE \"$_\" \"$response{$_}\"\n";
> checkresponse();
> }
> exit;
>
> sub checkresponse {
> my $input = <STDIN>;
> my @values;
>
> chomp $input;
> if ($input =~ /^200 result=(-?\d+)\s?(.*)$/) {
> warn "$name Command returned: $input\n" if ($debug);
> @values = ("$1", "$2");
> } else {
> $input .= <STDIN> if ($input =~ /^520-Invalid/);
> warn "$name Unexpected result: $input\n";
> @values = (-1, -1);
> }
> return @values;
> }
>
> sub detect_format {
> # Detect the sound format used #
> my @format;
> print "GET FULL VARIABLE \${CHANNEL(audionativeformat)}\n";
> my @reply = checkresponse();
> for ($reply[1]) {
> if (/(silk|sln)12/) { @format = ("sln12", 12000); }
> elsif (/(speex|slin|silk)16|g722|siren7/) { @format = ("sln16", 16000); }
> elsif (/(speex|slin|celt)32|siren14/) { @format = ("sln32", 32000); }
> elsif (/(celt|slin)44/) { @format = ("sln44", 44100); }
> elsif (/(celt|slin)48/) { @format = ("sln48", 48000); }
> else { @format = ("sln", 8000); }
> }
> return @format;
> }
>
> sub int_handler {
> die "$name Interrupt signal received, terminating...\n";
> }
>
> END {
> if ($tmpname) {
> warn "$name Cleaning temp files.\n" if ($debug);
> unlink glob "$tmpname.*";
> }
> }
>
> Atenciosamente
>
> Eng. Jose Antonio Sanchez
> Erimat Oeste Telecom
> www.pbxerix.com.br
> www.virtualpbxip.com.br
> www.erimatoeste.com.br
> jasanchez em terra.com.br
> 17-21393970 17-81437977
>
>
> Em 1 de fev de 2018, à(s) 14:44, Nuno Cunha <nuno.cunha em engdb.com.br>
> escreveu:
>
> Olas,
>
> Também tentei:
>
> p_channel.exec("Agi", "speech-recog.agi,pt-BR,2");
>
> com o mesmo resultado.
>
>
>
> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> Livre
> de vírus. www.avast.com
> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>.
>
>
>
> ----
> Nuno Correia dos Santos Cunha
>
> Analista de sistemas
> Automação & Controle - Serviços Logann - T&T
>
> Office:
> Skype:
>
> (31) 3211-7396
> nuno.cunha.engdb
> <http://www.engdb.com.br/> <https://www.facebook.com/engdb/>
> <https://www.linkedin.com/company-beta/27862?pathWildcard=27862>
> <https://www.youtube.com/channel/UCEUvHtjrY3-kD1tFryzCJyA> Avenida
> Getúlio Vargas, nº 1.300, 9º e 10º andares | Belo Horizonte – MG | CEP
> 30112-021
>
> Em 1 de fevereiro de 2018 14:23, Renato Santos <renato473 em gmail.com>
> escreveu:
>
>> Acho que está api foi descontinuada pois ela está na V1 e o Google usa a
>> V3
>>
>> Em qui, 1 de fev de 2018 às 13:11, Nuno Cunha <nuno.cunha em engdb.com.br>
>> escreveu:
>>
>>> Ola,
>>>
>>> Em meu AGI (fastAGI, em java) tenho estas duas linhas:
>>>
>>> p_channel.exec("Playback", "Fale o endereço!");
>>> p_channel.exec("Agi", "speech-recog.agi", "pt-BR");
>>>
>>>
>>> O problema é que após ler a pergunta o tempo de fala parece estar
>>> sendo de zero segundos, mal começo a falar e já recebo a mensagem que o
>>> endereço não foi compreendido (que é a lógica logo a seguir a estas duas
>>> linhas).
>>>
>>> No script speech-recog.agi tenho as configurações:
>>>
>>> # ----------------------------- #
>>> # User defined parameters: #
>>> # ----------------------------- #
>>> # Speech API key #
>>> my $key = "minha chave...";
>>>
>>> # Default language #
>>> my $language = "pt-BR";
>>>
>>> # Default max silence timeout #
>>> my $timeout = 2;
>>>
>>> # Absolute Recording timeout #
>>> my $abs_timeout = -1;
>>>
>>> # Default interrupt key #
>>> my $intkey = "#";
>>>
>>> # Input audio sample rate #
>>> # Leave blank to auto-detect #
>>> my $samplerate = "";
>>>
>>> # Profanity filter #
>>> my $pro_filter = "false";
>>>
>>> # Verbose debugging messages #
>>> my $debug = 0;
>>>
>>> # ----------------------------- #
>>>
>>>
>>> Testei colocando explicitamente na chamada do agi do Google o tempo de
>>> timeout, como mostrado abaixo, mas não mudou o comportamento.
>>>
>>> p_channel.exec("Playback", "Fale o endereço!");
>>> p_channel.exec("Agi", "speech-recog.agi", " pt-BR ", "2");
>>>
>>> Alguma ideia de como ter mais tempo para falar o endereço?
>>>
>>> Obrigado,
>>>
>>>
>>> ----
>>> Nuno Correia dos Santos Cunha
>>>
>>> Analista de sistemas
>>> Automação & Controle - Serviços Logann - T&T
>>>
>>> Office:
>>> Skype:
>>>
>>> (31) 3211-7396
>>> nuno.cunha.engdb
>>> <http://www.engdb.com.br/> <https://www.facebook.com/engdb/>
>>> <https://www.linkedin.com/company-beta/27862?pathWildcard=27862>
>>> <https://www.youtube.com/channel/UCEUvHtjrY3-kD1tFryzCJyA> Avenida
>>> Getúlio Vargas, nº 1.300, 9º e 10º andares | Belo Horizonte – MG | CEP
>>> 30112-021
>>>
>>>
>>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> Livre
>>> de vírus. www.avast.com
>>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>.
>>>
>>> _______________________________________________
>>> KHOMP: completa linha de placas externas FXO, FXS, GSM e E1
>>> Media Gateways de 1 a 64 E1s para SIP com R2, ISDN e SS7
>>> Intercomunicador e acesso remoto via rede IP e telefones IP
>>> Conheça todo o portfólio em www.Khomp.com <http://www.khomp.com/>
>>> _______________________________________________
>>> Para remover seu email desta lista, basta enviar um email em branco para
>>> asteriskbrasil-unsubscribe em listas.asteriskbrasil.org
>>
>> --
>> Enviado do Gmail para celular
>>
>> _______________________________________________
>> KHOMP: completa linha de placas externas FXO, FXS, GSM e E1
>> Media Gateways de 1 a 64 E1s para SIP com R2, ISDN e SS7
>> Intercomunicador e acesso remoto via rede IP e telefones IP
>> Conheça todo o portfólio em www.Khomp.com <http://www.khomp.com/>
>> _______________________________________________
>> Para remover seu email desta lista, basta enviar um email em branco para
>> asteriskbrasil-unsubscribe em listas.asteriskbrasil.org
>>
>
> _______________________________________________
> KHOMP: completa linha de placas externas FXO, FXS, GSM e E1
> Media Gateways de 1 a 64 E1s para SIP com R2, ISDN e SS7
> Intercomunicador e acesso remoto via rede IP e telefones IP
> Conheça todo o portfólio em www.Khomp.com
> _______________________________________________
> Para remover seu email desta lista, basta enviar um email em branco para
> asteriskbrasil-unsubscribe em listas.asteriskbrasil.org
>
>
>
> _______________________________________________
> KHOMP: completa linha de placas externas FXO, FXS, GSM e E1
> Media Gateways de 1 a 64 E1s para SIP com R2, ISDN e SS7
> Intercomunicador e acesso remoto via rede IP e telefones IP
> Conheça todo o portfólio em www.Khomp.com
> _______________________________________________
> Para remover seu email desta lista, basta enviar um email em branco para
> asteriskbrasil-unsubscribe em listas.asteriskbrasil.org
>
-------------- Próxima Parte ----------
Um anexo em HTML foi limpo...
URL: <http://asteriskbrasil.org/pipermail/asteriskbrasil/attachments/20180201/bed4dbff/attachment-0001.html>
-------------- Próxima Parte ----------
Um anexo não-texto foi limpo...
Nome: logo.erix.teste.fw.png
Tipo: image/png
Tamanho: 90365 bytes
Descrição: não disponível
URL: <http://asteriskbrasil.org/pipermail/asteriskbrasil/attachments/20180201/bed4dbff/attachment-0001.png>
Mais detalhes sobre a lista de discussão AsteriskBrasil