diff --git a/examples/common.cpp b/examples/common.cpp
index 99efd18d..9ab162a5 100644
--- a/examples/common.cpp
+++ b/examples/common.cpp
@@ -863,3 +863,21 @@ bool is_file_exist(const char *fileName)
     std::ifstream infile(fileName);
     return infile.good();
 }
+
+bool speak_with_file(const std::string & command, const std::string & text, const std::string & path, int voice_id)
+{
+    std::ofstream speak_file(path.c_str());
+    if (speak_file.fail()) {
+        fprintf(stderr, "%s: failed to open speak_file\n", __func__);
+        return false;
+    } else {
+        speak_file.write(text.c_str(), text.size());
+        speak_file.close();
+        int ret = system((command + " " + std::to_string(voice_id) + " " + path).c_str());
+        if (ret != 0) {
+            fprintf(stderr, "%s: failed to speak\n", __func__);
+            return false;
+        }
+    }
+    return true;
+}
diff --git a/examples/common.h b/examples/common.h
index 0529a057..2ed91ca9 100644
--- a/examples/common.h
+++ b/examples/common.h
@@ -306,3 +306,6 @@ int timestamp_to_sample(int64_t t, int n_samples, int whisper_sample_rate);
 
 // check if file exists using ifstream
 bool is_file_exist(const char *fileName);
+
+// write text to file, and call system("command voice_id file")
+bool speak_with_file(const std::string & command, const std::string & text, const std::string & path, int voice_id);
diff --git a/examples/talk-llama/.gitignore b/examples/talk-llama/.gitignore
index cbf36313..9c08e1f4 100644
--- a/examples/talk-llama/.gitignore
+++ b/examples/talk-llama/.gitignore
@@ -1 +1,2 @@
 audio.mp3
+to_speak.txt
diff --git a/examples/talk-llama/eleven-labs.py b/examples/talk-llama/eleven-labs.py
index edcd023b..7ed1d5dc 100644
--- a/examples/talk-llama/eleven-labs.py
+++ b/examples/talk-llama/eleven-labs.py
@@ -1,20 +1,80 @@
 import sys
-import importlib.util
+import argparse
+import textwrap
 
-if importlib.util.find_spec("elevenlabs") is None:
-    print("elevenlabs library is not installed, you can install it to your enviroment using 'pip install elevenlabs'")
+parser = argparse.ArgumentParser(add_help=False,
+    formatter_class=argparse.RawTextHelpFormatter)
+parser.add_argument("-q", "--quick", action="store_true",
+    help="skip checking the required library")
+
+modes = parser.add_argument_group("action")
+modes.add_argument("inputfile", metavar="TEXTFILE",
+    nargs='?', type=argparse.FileType(), default=sys.stdin,
+    help="read the text file (default: stdin)")
+modes.add_argument("-l", "--list", action="store_true",
+    help="show the list of voices and exit")
+modes.add_argument("-h", "--help", action="help",
+    help="show this help and exit")
+
+selopts = parser.add_argument_group("voice selection")
+selmodes = selopts.add_mutually_exclusive_group()
+selmodes.add_argument("-n", "--name",
+    default="Arnold",
+    help="get a voice object by name (default: Arnold)")
+selmodes.add_argument("-v", "--voice", type=int, metavar="NUMBER",
+    help="get a voice object by number (see --list)")
+selopts.add_argument("-f", "--filter", action="append", metavar="KEY=VAL",
+    default=["use case=narration"],
+    help=textwrap.dedent('''\
+        filter voices by labels (default: "use case=narration")
+        this option can be used multiple times
+        filtering will be disabled if the first -f has no "=" (e.g. -f "any")
+        '''))
+
+outmodes = parser.add_argument_group("output")
+outgroup = outmodes.add_mutually_exclusive_group()
+outgroup.add_argument("-s", "--save", metavar="FILE",
+    default="audio.mp3",
+    help="save the TTS to a file (default: audio.mp3)")
+outgroup.add_argument("-p", "--play", action="store_true",
+    help="play the TTS with ffplay")
+
+args = parser.parse_args()
+
+if not args.quick:
+    import importlib.util
+    if importlib.util.find_spec("elevenlabs") is None:
+        print("elevenlabs library is not installed, you can install it to your enviroment using 'pip install elevenlabs'")
+        sys.exit()
+
+from elevenlabs import voices, generate, play, save
+
+if args.filter and "=" in args.filter[0]:
+    voicelist = voices()
+    for f in args.filter:
+        label, value = f.split("=")
+        voicelist = filter(lambda x: x.labels.get(label) == value, voicelist)
+    voicelist = list(voicelist)
+else:
+    voicelist = list(voices())
+
+if args.list:
+    for i, v in enumerate(voicelist):
+        print(str(i) + ": " + v.name + " " + str(v.labels))
     sys.exit()
 
-from elevenlabs import generate, play, save
+if args.voice:
+    voice = voicelist[args.voice % len(voicelist)]
+else:
+    voice = args.name
+    # if -n should consult -f, use the following
+    #voice = next(x for x in voicelist if x.name == args.name)
 
-# Get a Voice object, by name or UUID
-voice = "Arnold" #Possible Voices: Adam Antoni Arnold Bella Domi Elli Josh
-
-# Generate the TTS
 audio = generate(
-  text=str(sys.argv[2:]),
-  voice=voice
+    text=str(args.inputfile.read()),
+    voice=voice
 )
-
-# Save the TTS to a file
-save(audio, "audio.mp3") 
+if args.play:
+    play(audio)
+else:
+    save(audio, args.save) 
diff --git a/examples/talk-llama/speak b/examples/talk-llama/speak
index 50e7210e..31ea417a 100755
--- a/examples/talk-llama/speak
+++ b/examples/talk-llama/speak
@@ -1,32 +1,40 @@
 #!/bin/bash
 
 # Usage:
-#  speak.sh <voice_id> <text-to-speak>
+#  speak <voice_id> <textfile>
 
-# espeak
-# Mac OS: brew install espeak
-# Linux: apt-get install espeak
-#
-#espeak -v en-us+m$1 -s 225 -p 50 -a 200 -g 5 -k 5 "$2"
+function installed() { command -v $1 >/dev/null 2>&1; }
 
-# piper
-#
-# https://github.com/rhasspy/piper
-#
-# Tested with Linux:
-#
-#echo "$2" | piper --model ~/en_US-lessac-medium.onnx --output-raw | aplay -q -r 22050 -f S16_LE -t raw -
+if installed espeak; then
+  espeak -v en-us+m$1 -s 225 -p 50 -a 200 -g 5 -k 5 -f $2
+
+elif installed piper && installed aplay; then
+  cat $2 | piper --model ~/en_US-lessac-medium.onnx --output-raw | aplay -q -r 22050 -f S16_LE -t raw -
 
 # for Mac
-say "$2"
+elif installed say; then
+  say -f $2
 
 # Eleven Labs
-# To use it, install the elevenlabs module from pip (pip install elevenlabs)
-# It's possible to use the API for free with limited number of characters. To increase this limit register to https://beta.elevenlabs.io to get an api key and paste it after 'ELEVEN_API_KEY='
-#Keep the line commented to use the free version whitout api key
-#
-#export ELEVEN_API_KEY=your_api_key
-#wd=$(dirname $0)
-#script=$wd/eleven-labs.py
-#python3 $script $1 "$2" >/dev/null 2>&1
-#ffplay -autoexit -nodisp -loglevel quiet -hide_banner -i ./audio.mp3 >/dev/null 2>&1
+elif installed python3 && \
+  python3 -c 'import importlib.util; exit(not importlib.util.find_spec("elevenlabs"))' && \
+  installed ffplay; then
+    # It's possible to use the API for free with limited number of characters.
+    # To increase this limit register to https://beta.elevenlabs.io to get an api key
+    # and paste it after 'ELEVEN_API_KEY='
+    # Keep the line commented to use the free version without api key
+    #export ELEVEN_API_KEY=your_api_key
+    wd=$(dirname $0)
+    script=$wd/eleven-labs.py
+    python3 $script -q -p -v $1 $2 >/dev/null 2>&1
+
+    # Uncomment to keep the audio file
+    #python3 $script -q -s ./audio.mp3 -v $1 $2 >/dev/null 2>&1
+    #ffplay -autoexit -nodisp -loglevel quiet -hide_banner -i ./audio.mp3 >/dev/null 2>&1
+
+else
+  echo 'Install espeak ("brew install espeak" or "apt-get install espeak"),'
+  echo 'piper ("pip install piper-tts" or https://github.com/rhasspy/piper) with aplay,'
+  echo 'or elevenlabs ("pip install elevenlabs") with ffplay.'
+  echo '(export ELEVEN_API_KEY if you have an api key from https://beta.elevenlabs.io)'
+fi
diff --git a/examples/talk-llama/speak.bat b/examples/talk-llama/speak.bat
index d719d690..91110c05 100644
--- a/examples/talk-llama/speak.bat
+++ b/examples/talk-llama/speak.bat
@@ -1 +1 @@
-@powershell -ExecutionPolicy Bypass -F examples\talk\speak.ps1 %1 %2
+@powershell -ExecutionPolicy Bypass -F examples\talk-llama\speak.ps1 %1 %2
diff --git a/examples/talk-llama/speak.ps1 b/examples/talk-llama/speak.ps1
index bdc4c5f8..51139586 100644
--- a/examples/talk-llama/speak.ps1
+++ b/examples/talk-llama/speak.ps1
@@ -1,12 +1,14 @@
 # Set-ExecutionPolicy -ExecutionPolicy Bypass -Scope CurrentUser
 param(
-  # voice options are David or Zira
-  [Parameter(Mandatory=$true)][string]$voice,
-  [Parameter(Mandatory=$true)][string]$text
+  [Parameter(Mandatory=$true)][int]$voicenum,
+  [Parameter(Mandatory=$true)][string]$textfile
 )
 
 Add-Type -AssemblyName System.Speech;
 $speak = New-Object System.Speech.Synthesis.SpeechSynthesizer;
-$speak.SelectVoice("Microsoft $voice Desktop");
+$voiceoptions = $speak.GetInstalledVoices("en-US");
+$voice = $voiceoptions[$voicenum % $voiceoptions.count];
+$speak.SelectVoice($voice.VoiceInfo.Name);
 $speak.Rate="0";
+$text = Get-Content -Path $textfile;
 $speak.Speak($text);
diff --git a/examples/talk-llama/talk-llama.cpp b/examples/talk-llama/talk-llama.cpp
index 60dd99e5..ddc9e765 100644
--- a/examples/talk-llama/talk-llama.cpp
+++ b/examples/talk-llama/talk-llama.cpp
@@ -75,6 +75,7 @@ struct whisper_params {
     std::string model_wsp   = "models/ggml-base.en.bin";
     std::string model_llama = "models/ggml-llama-7B.bin";
     std::string speak       = "./examples/talk-llama/speak";
+    std::string speak_file  = "./examples/talk-llama/to_speak.txt";
     std::string prompt      = "";
     std::string fname_out;
     std::string path_session = "";       // path to file for saving/loading model eval state
@@ -113,6 +114,7 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
         else if (arg == "-mw"  || arg == "--model-whisper")  { params.model_wsp      = argv[++i]; }
         else if (arg == "-ml"  || arg == "--model-llama")    { params.model_llama    = argv[++i]; }
         else if (arg == "-s"   || arg == "--speak")          { params.speak          = argv[++i]; }
+        else if (arg == "-sf"  || arg == "--speak-file")     { params.speak_file     = argv[++i]; }
         else if (arg == "--prompt-file")                     {
             std::ifstream file(argv[++i]);
             std::copy(std::istreambuf_iterator<char>(file), std::istreambuf_iterator<char>(), back_inserter(params.prompt));
@@ -160,6 +162,7 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
     fprintf(stderr, "  -mw FILE, --model-whisper  [%-7s] whisper model file\n",                          params.model_wsp.c_str());
     fprintf(stderr, "  -ml FILE, --model-llama    [%-7s] llama model file\n",                            params.model_llama.c_str());
     fprintf(stderr, "  -s FILE,  --speak TEXT     [%-7s] command for TTS\n",                             params.speak.c_str());
+    fprintf(stderr, "  -sf FILE, --speak-file     [%-7s] file to pass to TTS\n",                         params.speak_file.c_str());
     fprintf(stderr, "  --prompt-file FNAME        [%-7s] file with custom prompt to start dialog\n",     "");
     fprintf(stderr, "  --session FNAME                   file to cache model state in (may be large!) (default: none)\n");
     fprintf(stderr, "  -f FNAME, --file FNAME     [%-7s] text output file name\n",                       params.fname_out.c_str());
@@ -546,10 +549,7 @@ int main(int argc, char ** argv) {
 
                 // optionally give audio feedback that the current text is being processed
                 if (!params.heard_ok.empty()) {
-                    int ret = system((params.speak + " " + std::to_string(voice_id) + " '" + params.heard_ok + "'").c_str());
-                    if (ret != 0) {
-                        fprintf(stderr, "%s: failed to speak\n", __func__);
-                    }
+                    speak_with_file(params.speak, params.heard_ok, params.speak_file, voice_id);
                 }
 
                 // remove text between brackets using regex
@@ -748,11 +748,7 @@ int main(int argc, char ** argv) {
                     }
                 }
 
-                text_to_speak = ::replace(text_to_speak, "'", "'\"'\"'");
-                int ret = system((params.speak + " " + std::to_string(voice_id) + " '" + text_to_speak + "'").c_str());
-                if (ret != 0) {
-                    fprintf(stderr, "%s: failed to speak\n", __func__);
-                }
+                speak_with_file(params.speak, text_to_speak, params.speak_file, voice_id);
 
                 audio.clear();
             }
diff --git a/examples/talk/.gitignore b/examples/talk/.gitignore
index cbf36313..9c08e1f4 100644
--- a/examples/talk/.gitignore
+++ b/examples/talk/.gitignore
@@ -1 +1,2 @@
 audio.mp3
+to_speak.txt
diff --git a/examples/talk/eleven-labs.py b/examples/talk/eleven-labs.py
index edcd023b..7ed1d5dc 100644
--- a/examples/talk/eleven-labs.py
+++ b/examples/talk/eleven-labs.py
@@ -1,20 +1,80 @@
 import sys
-import importlib.util
+import argparse
+import textwrap
 
-if importlib.util.find_spec("elevenlabs") is None:
-    print("elevenlabs library is not installed, you can install it to your enviroment using 'pip install elevenlabs'")
+parser = argparse.ArgumentParser(add_help=False,
+    formatter_class=argparse.RawTextHelpFormatter)
+parser.add_argument("-q", "--quick", action="store_true",
+    help="skip checking the required library")
+
+modes = parser.add_argument_group("action")
+modes.add_argument("inputfile", metavar="TEXTFILE",
+    nargs='?', type=argparse.FileType(), default=sys.stdin,
+    help="read the text file (default: stdin)")
+modes.add_argument("-l", "--list", action="store_true",
+    help="show the list of voices and exit")
+modes.add_argument("-h", "--help", action="help",
+    help="show this help and exit")
+
+selopts = parser.add_argument_group("voice selection")
+selmodes = selopts.add_mutually_exclusive_group()
+selmodes.add_argument("-n", "--name",
+    default="Arnold",
+    help="get a voice object by name (default: Arnold)")
+selmodes.add_argument("-v", "--voice", type=int, metavar="NUMBER",
+    help="get a voice object by number (see --list)")
+selopts.add_argument("-f", "--filter", action="append", metavar="KEY=VAL",
+    default=["use case=narration"],
+    help=textwrap.dedent('''\
+        filter voices by labels (default: "use case=narration")
+        this option can be used multiple times
+        filtering will be disabled if the first -f has no "=" (e.g. -f "any")
+        '''))
+
+outmodes = parser.add_argument_group("output")
+outgroup = outmodes.add_mutually_exclusive_group()
+outgroup.add_argument("-s", "--save", metavar="FILE",
+    default="audio.mp3",
+    help="save the TTS to a file (default: audio.mp3)")
+outgroup.add_argument("-p", "--play", action="store_true",
+    help="play the TTS with ffplay")
+
+args = parser.parse_args()
+
+if not args.quick:
+    import importlib.util
+    if importlib.util.find_spec("elevenlabs") is None:
+        print("elevenlabs library is not installed, you can install it to your enviroment using 'pip install elevenlabs'")
+        sys.exit()
+
+from elevenlabs import voices, generate, play, save
+
+if args.filter and "=" in args.filter[0]:
+    voicelist = voices()
+    for f in args.filter:
+        label, value = f.split("=")
+        voicelist = filter(lambda x: x.labels.get(label) == value, voicelist)
+    voicelist = list(voicelist)
+else:
+    voicelist = list(voices())
+
+if args.list:
+    for i, v in enumerate(voicelist):
+        print(str(i) + ": " + v.name + " " + str(v.labels))
     sys.exit()
 
-from elevenlabs import generate, play, save
+if args.voice:
+    voice = voicelist[args.voice % len(voicelist)]
+else:
+    voice = args.name
+    # if -n should consult -f, use the following
+    #voice = next(x for x in voicelist if x.name == args.name)
 
-# Get a Voice object, by name or UUID
-voice = "Arnold" #Possible Voices: Adam Antoni Arnold Bella Domi Elli Josh
-
-# Generate the TTS
 audio = generate(
-  text=str(sys.argv[2:]),
-  voice=voice
+    text=str(args.inputfile.read()),
+    voice=voice
 )
-
-# Save the TTS to a file
-save(audio, "audio.mp3") 
+if args.play:
+    play(audio)
+else:
+    save(audio, args.save) 
diff --git a/examples/talk/speak b/examples/talk/speak
index b822f615..31ea417a 100644
--- a/examples/talk/speak
+++ b/examples/talk/speak
@@ -1,24 +1,40 @@
 #!/bin/bash
 
 # Usage:
-#  speak.sh <voice_id> <text-to-speak>
+#  speak <voice_id> <textfile>
 
-# espeak
-# Mac OS: brew install espeak
-# Linux: apt-get install espeak
-#
-#espeak -v en-us+m$1 -s 175 -p 50 -a 200 -g 5 -k 5 "$2"
+function installed() { command -v $1 >/dev/null 2>&1; }
 
-# Mac OS "say" command
-say "$2"
+if installed espeak; then
+  espeak -v en-us+m$1 -s 225 -p 50 -a 200 -g 5 -k 5 -f $2
+
+elif installed piper && installed aplay; then
+  cat $2 | piper --model ~/en_US-lessac-medium.onnx --output-raw | aplay -q -r 22050 -f S16_LE -t raw -
+
+# for Mac
+elif installed say; then
+  say -f $2
 
 # Eleven Labs
-# To use it, install the elevenlabs module from pip (pip install elevenlabs)
-# It's possible to use the API for free with limited number of characters. To increase this limit register to https://beta.elevenlabs.io to get an api key and paste it after 'ELEVEN_API_KEY='
-#Keep the line commented to use the free version without api key
-#
-#export ELEVEN_API_KEY=your_api_key
-#wd=$(dirname $0)
-#script=$wd/eleven-labs.py
-#python3 $script $1 "$2"
-#ffplay -autoexit -nodisp -loglevel quiet -hide_banner -i ./audio.mp3
+elif installed python3 && \
+  python3 -c 'import importlib.util; exit(not importlib.util.find_spec("elevenlabs"))' && \
+  installed ffplay; then
+    # It's possible to use the API for free with limited number of characters.
+    # To increase this limit register to https://beta.elevenlabs.io to get an api key
+    # and paste it after 'ELEVEN_API_KEY='
+    # Keep the line commented to use the free version without api key
+    #export ELEVEN_API_KEY=your_api_key
+    wd=$(dirname $0)
+    script=$wd/eleven-labs.py
+    python3 $script -q -p -v $1 $2 >/dev/null 2>&1
+
+    # Uncomment to keep the audio file
+    #python3 $script -q -s ./audio.mp3 -v $1 $2 >/dev/null 2>&1
+    #ffplay -autoexit -nodisp -loglevel quiet -hide_banner -i ./audio.mp3 >/dev/null 2>&1
+
+else
+  echo 'Install espeak ("brew install espeak" or "apt-get install espeak"),'
+  echo 'piper ("pip install piper-tts" or https://github.com/rhasspy/piper) with aplay,'
+  echo 'or elevenlabs ("pip install elevenlabs") with ffplay.'
+  echo '(export ELEVEN_API_KEY if you have an api key from https://beta.elevenlabs.io)'
+fi
diff --git a/examples/talk/speak.ps1 b/examples/talk/speak.ps1
index bdc4c5f8..51139586 100644
--- a/examples/talk/speak.ps1
+++ b/examples/talk/speak.ps1
@@ -1,12 +1,14 @@
 # Set-ExecutionPolicy -ExecutionPolicy Bypass -Scope CurrentUser
 param(
-  # voice options are David or Zira
-  [Parameter(Mandatory=$true)][string]$voice,
-  [Parameter(Mandatory=$true)][string]$text
+  [Parameter(Mandatory=$true)][int]$voicenum,
+  [Parameter(Mandatory=$true)][string]$textfile
 )
 
 Add-Type -AssemblyName System.Speech;
 $speak = New-Object System.Speech.Synthesis.SpeechSynthesizer;
-$speak.SelectVoice("Microsoft $voice Desktop");
+$voiceoptions = $speak.GetInstalledVoices("en-US");
+$voice = $voiceoptions[$voicenum % $voiceoptions.count];
+$speak.SelectVoice($voice.VoiceInfo.Name);
 $speak.Rate="0";
+$text = Get-Content -Path $textfile;
 $speak.Speak($text);
diff --git a/examples/talk/talk.cpp b/examples/talk/talk.cpp
index f9de3048..c1c6f8ba 100644
--- a/examples/talk/talk.cpp
+++ b/examples/talk/talk.cpp
@@ -38,6 +38,7 @@ struct whisper_params {
     std::string model_wsp = "models/ggml-base.en.bin";
     std::string model_gpt = "models/ggml-gpt-2-117M.bin";
     std::string speak     = "./examples/talk/speak";
+    std::string speak_file= "./examples/talk/to_speak.txt";
     std::string fname_out;
 };
 
@@ -68,6 +69,7 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
         else if (arg == "-mw"  || arg == "--model-whisper") { params.model_wsp     = argv[++i]; }
         else if (arg == "-mg"  || arg == "--model-gpt")     { params.model_gpt     = argv[++i]; }
         else if (arg == "-s"   || arg == "--speak")         { params.speak         = argv[++i]; }
+        else if (arg == "-sf"  || arg == "--speak_file")    { params.speak_file    = argv[++i]; }
         else if (arg == "-f"   || arg == "--file")          { params.fname_out     = argv[++i]; }
         else {
             fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
@@ -102,6 +104,7 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
     fprintf(stderr, "  -mw FILE, --model-whisper [%-7s] whisper model file\n",                          params.model_wsp.c_str());
     fprintf(stderr, "  -mg FILE, --model-gpt     [%-7s] gpt model file\n",                              params.model_gpt.c_str());
     fprintf(stderr, "  -s FILE,  --speak TEXT    [%-7s] command for TTS\n",                             params.speak.c_str());
+    fprintf(stderr, "  -sf FILE, --speak_file    [%-7s] file to pass to TTS\n",                         params.speak_file.c_str());
     fprintf(stderr, "  -f FNAME, --file FNAME    [%-7s] text output file name\n",                       params.fname_out.c_str());
     fprintf(stderr, "\n");
 }
@@ -316,7 +319,7 @@ int main(int argc, char ** argv) {
                     std::string prompt = ::replace(::replace(k_prompt, "{0}", params.person), "{1}", prompt_base);
 
                     text_to_speak = gpt2_gen_text(ctx_gpt, prompt.c_str(), params.max_tokens);
-                    text_to_speak = std::regex_replace(text_to_speak, std::regex("[^a-zA-Z0-9\\.,\\?!\\s\\:\\'\\-]"), "");
+                    //text_to_speak = std::regex_replace(text_to_speak, std::regex("[^a-zA-Z0-9\\.,\\?!\\s\\:\\'\\-]"), "");
                     text_to_speak = text_to_speak.substr(0, text_to_speak.find_first_of('\n'));
 
                     // remove first 2 lines of base prompt
@@ -354,10 +357,7 @@ int main(int argc, char ** argv) {
                 gpt2_set_prompt(ctx_gpt, prompt_base.c_str());
 
                 text_to_speak = ::replace(text_to_speak, params.person + ": ", "");
-                int ret = system((params.speak + " " + std::to_string(voice_id) + " \"" + text_to_speak + "\"").c_str());
-                if (ret != 0) {
-                    fprintf(stderr, "%s: system() failed!\n", __func__);
-                }
+                speak_with_file(params.speak, text_to_speak, params.speak_file, voice_id);
 
                 audio.clear();