speech-recognition: iterating the backend/notify API.

diff --git a/src/daemon/client.h b/src/daemon/client.h
index 3ee4805..154c2af 100644 (file)
--- a/src/daemon/client.h
+++ b/src/daemon/client.h
@@ -99,9 +99,9 @@ struct srs_client_s {
 
 /** Create a new client. */
 srs_client_t *client_create(srs_context_t *srs, srs_client_type_t type,
-                        const char *name, const char *appclass,
-                        char **commands, int ncommand,
-                        const char *id, srs_client_ops_t *ops);
+                            const char *name, const char *appclass,
+                            char **commands, int ncommand,
+                            const char *id, srs_client_ops_t *ops);
 
 /** Destroy a client. */
 void client_destroy(srs_client_t *c);

diff --git a/src/daemon/config.c b/src/daemon/config.c
index 7ebf061..735d26e 100644 (file)
--- a/src/daemon/config.c
+++ b/src/daemon/config.c
@@ -67,6 +67,8 @@ static void print_usage(const char *argv0, int exit_code, const char *fmt, ...)
            "The possible options are:\n"
            "  -c, --config-file=PATH         main configuration file to use\n"
            "      The default configuration file is '%s'.\n"
+           "  -B, --dbus=BUS                 D-BUS type or address to use\n"
+           "      The default D-BUS is 'session'.\n"
            "  -P, --plugin-dir=PATH          use DIR to search for plugins\n"
            "      The default plugin directory is '%s'.\n"
            "  -L, --load-plugin=NAME         load the given plugin\n"

diff --git a/src/daemon/recognizer.c b/src/daemon/recognizer.c
index d29fbe4..0c0730b 100644 (file)
--- a/src/daemon/recognizer.c
+++ b/src/daemon/recognizer.c
@@ -42,8 +42,7 @@ typedef struct {
 } srs_srec_t;
 
 static srs_srec_t *find_srec(srs_context_t *srs, const char *name);
-static void srec_notify_cb(srs_srec_token_t *tokens, int ntoken,
-                           void *notify_data);
+static int srec_notify_cb(srs_srec_utterance_t *utt, void *notify_data);
 
 int srs_register_srec(srs_context_t *srs, const char *name,
                       srs_srec_api_t *api, void *api_data,
@@ -131,47 +130,24 @@ void srs_deactivate_srec(srs_context_t *srs, const char *name)
 }
 
 
-int srs_check_model(srs_context_t *srs, const char *name, const char *model)
+int srs_check_decoder(srs_context_t *srs, const char *name,
+                      const char *decoder)
 {
     srs_srec_t *srec = find_srec(srs, name);
 
     if (srec != NULL)
-        return srec->api.check_model(model, srec->api_data);
+        return srec->api.check_decoder(decoder, srec->api_data);
     else
         return FALSE;
 }
 
 
-int srs_check_dictionary(srs_context_t *srs, const char *name,
-                         const char *dictionary)
+int srs_set_decoder(srs_context_t *srs, const char *name, const char *decoder)
 {
     srs_srec_t *srec = find_srec(srs, name);
 
     if (srec != NULL)
-        return srec->api.check_dictionary(dictionary, srec->api_data);
-    else
-        return FALSE;
-}
-
-
-int srs_set_model(srs_context_t *srs, const char *name, const char *model)
-{
-    srs_srec_t *srec = find_srec(srs, name);
-
-    if (srec != NULL)
-        return srec->api.set_model(model, srec->api_data);
-    else
-        return FALSE;
-}
-
-
-int srs_set_dictionary(srs_context_t *srs, const char *name,
-                       const char *dictionary)
-{
-    srs_srec_t *srec = find_srec(srs, name);
-
-    if (srec != NULL)
-        return srec->api.set_dictionary(dictionary, srec->api_data);
+        return srec->api.select_decoder(decoder, srec->api_data);
     else
         return FALSE;
 }
@@ -199,14 +175,22 @@ static srs_srec_t *find_srec(srs_context_t *srs, const char *name)
 }
 
 
-static void srec_notify_cb(srs_srec_token_t *tokens, int ntoken,
-                           void *notify_data)
+static int srec_notify_cb(srs_srec_utterance_t *utt, void *notify_data)
 {
-    srs_srec_t *srec = (srs_srec_t *)notify_data;
-    int         i;
-
-    mrp_log_info("Got %d tokens from %s backend:", ntoken, srec->name);
+    srs_srec_t           *srec = (srs_srec_t *)notify_data;
+    srs_srec_candidate_t *c;
+    srs_srec_token_t     *t;
+    int                   i, j;
+
+    mrp_log_info("Got %zd recognition candidates in from %s backend:",
+                 utt->ncand, srec->name);
+
+    for (i = 0, c = utt->cands; i < (int)utt->ncand; i++, c++) {
+        mrp_log_info("Candidate #%d:", i);
+        for (j = 0, t = c->tokens; j < (int)c->ntoken; j++, t++) {
+            mrp_log_info("    token #%d: '%s'", j, t->token);
+        }
+    }
 
-    for (i = 0; i < ntoken; i++)
-        mrp_log_info("    #%d: '%s'", i, tokens[i].token);
+    return -1;
 }

diff --git a/src/daemon/recognizer.h b/src/daemon/recognizer.h
index b9c693c..faea78d 100644 (file)
--- a/src/daemon/recognizer.h
+++ b/src/daemon/recognizer.h
@@ -31,11 +31,10 @@
 #define __SRS_DAEMON_RECOGNIZER_H__
 
 /** Type for tokens recognized by a speech recognition backend. */
-typedef struct srs_srec_token_s srs_srec_token_t;
+typedef struct srs_srec_utterance_s srs_srec_utterance_t;
 
 /** Type for a backend recognition notification callback. */
-typedef void (*srs_srec_notify_t)(srs_srec_token_t *tokens, int ntoken,
-                                  void *notify_data);
+typedef int (*srs_srec_notify_t)(srs_srec_utterance_t *utt, void *notify_data);
 
 /*
  * API to a speech recognition backend.
@@ -52,27 +51,43 @@ typedef struct {
     /** Get a copy of the audio samples in the buffer. */
     void *(*sampledup)(uint32_t start, uint32_t end, void *user_data);
     /** Check if the given language model exists/is usable. */
-    int (*check_model)(const char *model, void *user_data);
-    /** Check if the given dictionary exists/is usable. */
-    int (*check_dictionary)(const char *dictionary, void *user_data);
+    int (*check_decoder)(const char *decoder, void *user_data);
     /** Set language model to be used. */
-    int (*set_model)(const char *model, void *user_data);
-    /** Set dictionary to be used. */
-    int (*set_dictionary)(const char *dictionary, void *user_data);
+    int (*select_decoder)(const char *decoder, void *user_data);
 } srs_srec_api_t;
 
 /*
  * a single speech token
  */
-struct srs_srec_token_s {
+typedef struct {
     char     *token;                     /* recognized tokens */
     double    score;                     /* correctness probability */
     uint32_t  start;                     /* start in audio buffer */
     uint32_t  end;                       /* end in audio buffer */
-    int       flush : 1;                 /* flush from audio buffer */
-};
+} srs_srec_token_t;
+
+
+/*
+ * a single candidate
+ */
+typedef struct {
+    double            score;
+    size_t            ntoken;
+    srs_srec_token_t *tokens;
+} srs_srec_candidate_t;
 
 
+/*
+ * an utterance with a set of candidates
+ */
+struct srs_srec_utterance_s {
+    const char           *id;
+    double                score;
+    uint32_t              length;
+    size_t                ncand;
+    srs_srec_candidate_t *cands;
+};
+
 /** Register a speech recognition backend. */
 int srs_register_srec(srs_context_t *srs, const char *name,
                       srs_srec_api_t *api, void *api_data,
@@ -87,18 +102,11 @@ int srs_activate_srec(srs_context_t *srs, const char *name);
 /** Deactivate the specified speech recognition backend. */
 void srs_deactivate_srec(srs_context_t *srs, const char *name);
 
-/** Check if a given model exists for a recognition backend. */
-int srs_check_model(srs_context_t *srs, const char *name, const char *model);
-
-/** Check if a given dictionary exists for a recognition backend. */
-int srs_check_dictionary(srs_context_t *srs, const char *name,
-                          const char *dictionary);
-
-/** Check if a given model exists for a recognition backend. */
-int srs_set_model(srs_context_t *srs, const char *name, const char *model);
+/** Check if a decoder (model/dictionary combination) exists for a backend. */
+int srs_check_decoder(srs_context_t *srs, const char *name,
+                      const char *decoder);
 
-/** Check if a given dictionary exists for a recognition backend. */
-int srs_set_dictionary(srs_context_t *srs, const char *name,
-                       const char *dictionary);
+/** Select a decoder for a backend. */
+int srs_set_decoder(srs_context_t *srs, const char *name, const char *decoder);
 
 #endif /* __SRS_DAEMON_RECOGNIZER_H__ */

diff --git a/src/plugins/fake-speech-engine/fake.c b/src/plugins/fake-speech-engine/fake.c
index 9ac782d..346d290 100644 (file)
--- a/src/plugins/fake-speech-engine/fake.c
+++ b/src/plugins/fake-speech-engine/fake.c
@@ -87,9 +87,11 @@ static int arm_token_timer(fake_t *fake, double delay)
 
 static void push_token_cb(mrp_timer_t *t, void *user_data)
 {
-    fake_t           *fake  = (fake_t *)user_data;
-    fake_token_t     *token = fake->tokens + fake->tokidx++;
-    srs_srec_token_t  tok;
+    fake_t               *fake  = (fake_t *)user_data;
+    fake_token_t         *token = fake->tokens + fake->tokidx++;
+    srs_srec_token_t      tok;
+    srs_srec_candidate_t  cand;
+    srs_srec_utterance_t  utt;
 
     mrp_del_timer(t);
     fake->toktmr = NULL;
@@ -105,9 +107,19 @@ static void push_token_cb(mrp_timer_t *t, void *user_data)
     tok.score = 1;
     tok.start = token - fake->tokens;
     tok.end   = tok.start + 1;
-    tok.flush = FALSE;
 
-    fake->notify(&tok, 1, fake->notify_data);
+    cand.score  = 1;
+    cand.ntoken = 1;
+    cand.tokens = &tok;
+
+    utt.id     = "fake backend utterance";
+    utt.score  = 1;
+    utt.length = 1;
+    utt.length = tok.end - tok.start;
+    utt.ncand  = 1;
+    utt.cands  = &cand;
+
+    fake->notify(&utt, fake->notify_data);
 }
 
 
@@ -186,49 +198,25 @@ static void *fake_sampledup(uint32_t start, uint32_t end, void *user_data)
 }
 
 
-static int fake_check_model(const char *model, void *user_data)
-{
-    fake_t *fake = (fake_t *)user_data;
-
-    MRP_UNUSED(fake);
-
-    mrp_debug("checking model '%s' for fake backend", model);
-
-    return TRUE;
-}
-
-
-static int fake_check_dictionary(const char *dictionary, void *user_data)
-{
-    fake_t *fake = (fake_t *)user_data;
-
-    MRP_UNUSED(fake);
-
-    mrp_debug("checking dictionary '%s' for fake backend", dictionary);
-
-    return TRUE;
-}
-
-
-static int fake_set_model(const char *model, void *user_data)
+static int fake_check_decoder(const char *decoder, void *user_data)
 {
     fake_t *fake = (fake_t *)user_data;
 
     MRP_UNUSED(fake);
 
-    mrp_debug("setting model '%s' for fake backend", model);
+    mrp_debug("checking availibilty of decoder '%s' for fake backend", decoder);
 
     return TRUE;
 }
 
 
-static int fake_set_dictionary(const char *dictionary, void *user_data)
+static int fake_select_decoder(const char *decoder, void *user_data)
 {
     fake_t *fake = (fake_t *)user_data;
 
     MRP_UNUSED(fake);
 
-    mrp_debug("setting dictionary '%s' for fake backend", dictionary);
+    mrp_debug("selecting decoder '%s' for fake backend", decoder);
 
     return TRUE;
 }
@@ -242,10 +230,8 @@ static int create_fake(srs_plugin_t *plugin)
     flush:            fake_flush,
     rescan:           fake_rescan,
     sampledup:        fake_sampledup,
-    check_model:      fake_check_model,
-    check_dictionary: fake_check_dictionary,
-    set_model:        fake_set_model,
-    set_dictionary:   fake_set_dictionary,
+    check_decoder:    fake_check_decoder,
+    select_decoder:   fake_select_decoder,
     };
 
     srs_context_t *srs = plugin->srs;

diff --git a/src/plugins/nuance-speech-engine/nuance.c b/src/plugins/nuance-speech-engine/nuance.c
index a62ae08..0d95690 100644 (file)
--- a/src/plugins/nuance-speech-engine/nuance.c
+++ b/src/plugins/nuance-speech-engine/nuance.c
@@ -109,49 +109,26 @@ static void *nuance_sampledup(uint32_t start, uint32_t end, void *user_data)
 }
 
 
-static int nuance_check_model(const char *model, void *user_data)
+static int nuance_check_decoder(const char *decoder, void *user_data)
 {
     nuance_t *nua = (nuance_t *)user_data;
 
     MRP_UNUSED(nua);
 
-    mrp_debug("checking model '%s' for Nuance backend", model);
+    mrp_debug("checking availability of decoder '%s' for Nuance backend",
+              decoder);
 
     return TRUE;
 }
 
 
-static int nuance_check_dictionary(const char *dictionary, void *user_data)
+static int nuance_select_decoder(const char *decoder, void *user_data)
 {
     nuance_t *nua = (nuance_t *)user_data;
 
     MRP_UNUSED(nua);
 
-    mrp_debug("checking dictionary '%s' for Nuance backend", dictionary);
-
-    return TRUE;
-}
-
-
-static int nuance_set_model(const char *model, void *user_data)
-{
-    nuance_t *nua = (nuance_t *)user_data;
-
-    MRP_UNUSED(nua);
-
-    mrp_debug("setting model '%s' for Nuance backend", model);
-
-    return TRUE;
-}
-
-
-static int nuance_set_dictionary(const char *dictionary, void *user_data)
-{
-    nuance_t *nua = (nuance_t *)user_data;
-
-    MRP_UNUSED(nua);
-
-    mrp_debug("setting dictionary '%s' for Nuance backend", dictionary);
+    mrp_debug("setting decoder '%s' for Nuance backend", decoder);
 
     return TRUE;
 }
@@ -165,10 +142,8 @@ static int create_nuance(srs_plugin_t *plugin)
     flush:            nuance_flush,
     rescan:           nuance_rescan,
     sampledup:        nuance_sampledup,
-    check_model:      nuance_check_model,
-    check_dictionary: nuance_check_dictionary,
-    set_model:        nuance_set_model,
-    set_dictionary:   nuance_set_dictionary,
+    check_decoder:    nuance_check_decoder,
+    select_decoder:   nuance_select_decoder,
     };
 
     srs_context_t *srs = plugin->srs;