{"id":4439,"date":"2022-02-16T20:02:46","date_gmt":"2022-02-16T19:02:46","guid":{"rendered":"https:\/\/www.pschatzmann.ch\/home\/?p=4439"},"modified":"2022-02-25T11:19:56","modified_gmt":"2022-02-25T10:19:56","slug":"tts-with-prerecorded-audio-building-a-talking-clock","status":"publish","type":"post","link":"https:\/\/www.pschatzmann.ch\/home\/2022\/02\/16\/tts-with-prerecorded-audio-building-a-talking-clock\/","title":{"rendered":"TTS with Pre-Recorded Audio: Building a Talking Clock"},"content":{"rendered":"

Last year I was digging into Arduino Based TTS Solutions and came to the conclusion that the available engines will not provide any quality audio<\/a> and therefore recommended to consider an approach which is based on recorded audio samples<\/strong>.<\/p>\n

Today I took the opportunity to create the new arduino-simple-tts<\/a> project which is based on this approach. As a proof of concept I have implemented the speaking of numbers<\/strong> and of the output of time<\/strong>. All the relevant words have been recorded as mp3 files<\/strong> and are stored in program memory<\/strong>.<\/p>\n

Arduino Sketch<\/h2>\n

Here is a Example Sketch which implements a Speaking Clock:<\/p>\n

#include \"SimpleTTS.h\"\n#include \"AudioCodecs\/CodecMP3Helix.h\"\n#include \"AudioLibs\/AudioKit.h\"\n#include \"TimeInfo.h\"\n\n\/\/ Output\nTimeToText ttt;\nAudioKitStream i2s; \/\/ replace with alterntive Audio Sink if needed: AnalogAudioStream, I2SStream etc.\nMP3DecoderHelix mp3;\nAudioDictionary dictionary(ExampleAudioDictionaryValues);\nTextToSpeech tts(ttt, i2s, mp3, dictionary);\n\n\/\/ Determine Time\nTimeInfo timeInfo;\nconst char* ssid = \"SSID\";\nconst char* password = \"password\";\n\nvoid setup() {\n  Serial.begin(115200);\n  AudioLogger::instance().begin(Serial, AudioLogger::Info);\n\n  \/\/ setup i2s\n  auto cfg = i2s.defaultConfig();\n  cfg.sample_rate = 24000;\n  cfg.channels = 1;\n  i2s.begin(cfg);\n\n  \/\/ We announce the time only every 5 minutes\n  timeInfo.setEveryMinutes(5);\n  \/\/ start WIFI and time\n  timeInfo.begin(ssid, password);\n  ttt.say(timeInfo.time());\n}\n\nvoid loop() {\n  \/\/ speach output\n  if (timeInfo.update()){\n    ttt.say(timeInfo.time());\n  }\n}\n<\/code><\/pre>\n
The TimeToText class<\/strong> is translating the time into words which is the input to the TextToSpeech class<\/strong> which handles the audio. This class is based on my Arduino Audio Tools<\/a> library: so we need to feed it with a OutputSink and a MP3 Decoder. The audio samples are determined with the help of the AudioDictionary<\/strong>. As part of the sketch I have implemented the TimeInfo<\/strong> class which just retrieves the time information from a time server and determines if we need to announce a new time.<\/p>\n
The full source code<\/a> is available on Github<\/p>\n
Memory Requirements<\/h2>\n
The sketch which includes the audio data is only using 37% of the program storage:<\/p>\n
Sketch uses 1171698 bytes (37%) of program storage space. Maximum is 3145728 bytes.\nGlobal variables use 48232 bytes (14%) of dynamic memory, leaving 279448 bytes for \n<\/code><\/pre>\n
I think this is quite impressive and we have quite some headroom before we need to resort to the samples being stored on a SD drive.<\/p>\n
Next Steps<\/h2>\n
I see three things that could be improved:<\/p>\n