diff --git a/README.md b/README.md
index 6409c03..a4d6130 100644
--- a/README.md
+++ b/README.md
@@ -15,18 +15,7 @@ Language-specific build instructions can be found in their respective directorie
 Techmo TTS Service API is defined in `proto/TTS.proto` file.
 
 Service's `Synthesize` method accepts `SynthesizeRequest` object which contains whole phrase to be synthesized.
-You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form. In that string you can use several special tags which can be interpreted. Tags have to be in from `<tag>something special</tag>` and can occur in any place in text. Currently interpreted tags are:
-
-cardinal    cardinal number     "<cardinal>7</cardinal>"    -> "siedem"
-signed      number with sign    "<signed>-15</signed>"      -> "minus piętnaście"
-ordinal     ordinal number      "<ordinal>1</ordinal>"      -> "pierwszy"
-fraction    fractional number   "<fraction>3/4</fraction>"  -> "trzy czwarte"
-postal      postal code         "<postal>30-020</postal>"   -> "trzydzieści zero dwadzieścia"
-time        time                "<time>22:00</time>"        -> "dwudziesta druga"
-date        date                "<date>12/05/2001</date>"   -> "dwunasty maja dwa tysiące jeden"
-
-Note: when interpreting tags only nominal case is supported at the moment.
-
+You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form.
 You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. Currently supported option is only `sample_rate_hertz`, which is desired sampling frequency (in hertz) of synthesized audio.
 
 `SynthesizeRequest` can be sent to the service via gRPC insecure channel (that does not require authentication).
@@ -37,3 +26,10 @@ We provide sample TTS Client written in:
 - C++ in `cpp` (accepts text to be synthesized as a command line string),  
 - Python in `python` (accepts text to be synthesized as a command line string or as a content of given text file).  
 By default it saves "TechmoTTS.wav" file with received synthesized audio content. To use it, you have to specify provided by us service IP address and port using `--service-address` option with string in form "address:port".
+
+TTS supports following SSML tags:  
+`<prosody volume="+40%">text</prosody>`	Change of volume.			Parses numeric value (+ and -)  
+`<prosody rate="-10%">text</prosody>`	Change of rate without change of pitch.	Parses numeric value (+ and -)  
+`<prosody pitch="+30%">text</prosody>`	Change of pitch.			Parses numeric value (+ and -)  
+`<break time="3s"/>`	Inserts silence.					Parses numeric value in seconds or milliseconds ("3s", "750ms")  
+Prosody tags can be combined together, example: `<prosody rate="+30%" pitch="+40%">text</prosody>`.
diff --git a/cpp/CHANGELOG.md b/cpp/CHANGELOG.md
index 33273cc..0f54eff 100644
--- a/cpp/CHANGELOG.md
+++ b/cpp/CHANGELOG.md
@@ -1,5 +1,9 @@
 # Tribune TTS gRPC C++ client Changelog
 
+## [1.3.0] - 2019-06-24
+### Added
+- `audio-encoding` option.
+
 ## [1.2.0] - 2018-12-12
 ### Added
 - Support for setting gRPC deadline (how long the client is willing to wait for a reply from the server).
diff --git a/cpp/libtribune-client/VERSION.h b/cpp/libtribune-client/VERSION.h
index 7da2f5f..4b22c6d 100644
--- a/cpp/libtribune-client/VERSION.h
+++ b/cpp/libtribune-client/VERSION.h
@@ -1 +1 @@
-static constexpr auto LIBTRIBUNE_CLIENT_VERSION = "1.2.0";
+static constexpr auto LIBTRIBUNE_CLIENT_VERSION = "1.3.0";
diff --git a/cpp/libtribune-client/tribune_client.cpp b/cpp/libtribune-client/tribune_client.cpp
index f1f0793..c2ef5ba 100644
--- a/cpp/libtribune-client/tribune_client.cpp
+++ b/cpp/libtribune-client/tribune_client.cpp
@@ -1,16 +1,20 @@
 #include <grpc++/grpc++.h>
 #include <google/protobuf/text_format.h>
 
-#include "tribune_tts.grpc.pb.h"
 #include "tribune_client.h"
 
 
 namespace techmo { namespace tribune {
 
 SynthesizeRequest build_request(const TribuneClientConfig& config, const std::string& text) {
+    if(config.encoding == AudioEncoding::OGG_OPUS and config.sample_rate_hertz != 0 ){
+        throw std::runtime_error("Custom sample rate is not supported with Opus compression.");
+    }
+
     SynthesizeRequest request;
     request.set_text(text);
     request.mutable_config()->set_sample_rate_hertz(config.sample_rate_hertz);
+    request.mutable_config()->set_encoding(config.encoding);
     return request;
 }
 
diff --git a/cpp/libtribune-client/tribune_client.h b/cpp/libtribune-client/tribune_client.h
index 72dfe0a..2cdd773 100644
--- a/cpp/libtribune-client/tribune_client.h
+++ b/cpp/libtribune-client/tribune_client.h
@@ -1,6 +1,7 @@
 #ifndef __TRIBUNE_CLIENT_H__
 #define __TRIBUNE_CLIENT_H__
 
+#include "tribune_tts.grpc.pb.h"
 
 namespace techmo { namespace tribune {
 
@@ -9,6 +10,7 @@ struct TribuneClientConfig {
                                             // Session ID is the best way to match log's from client application with these on server side.
     int grpc_timeout = 0;                   // Timeout in milliseconds used to set gRPC deadline - how long the client is willing to wait for a reply from the server.
     unsigned int sample_rate_hertz = 0;     // Sample rate in Hz of synthesized audio. If set to 0, the service will use voice's original sample rate.
+    AudioEncoding encoding = AudioEncoding::LINEAR16;
 };
 
 struct TribuneAudioData {
diff --git a/cpp/libtribune-client/tribune_tts.grpc.pb.h b/cpp/libtribune-client/tribune_tts.grpc.pb.h
index 9fed3d8..48ff534 100644
--- a/cpp/libtribune-client/tribune_tts.grpc.pb.h
+++ b/cpp/libtribune-client/tribune_tts.grpc.pb.h
@@ -3,9 +3,9 @@
 // source: tribune_tts.proto
 // Original file comments:
 // Techmo Tribune TTS API  
-// version: 1.1.0   
-// authors: Dawid Skurzok, Paweł Jaciów  
-// date:    2018-01-25  
+// version: 1.2.0   
+// authors: Dawid Skurzok, Paweł Jaciów, Jerzy Skrzypek
+// date:    2019-06-25  
 #ifndef GRPC_tribune_5ftts_2eproto__INCLUDED
 #define GRPC_tribune_5ftts_2eproto__INCLUDED
 
@@ -32,28 +32,24 @@ class ServerContext;
 namespace techmo {
 namespace tribune {
 
-// Service that implements Techmo Text-To-Speech (TTS) API.
+// Techmo Text-To-Speech (TTS) API.
 //
 // Service's `Synthesize` method accepts `SynthesizeRequest` object which contains whole phrase to be synthesized.  
-// You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form. In that string you can use several special tags which can be interpreted. Tags have to be in from `<tag>something special</tag>` and can occur in any place in text. Currently interpreted tags are:
+// You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form.
 //
-// | Tag | Description | Example (input) | Example (output) |
-// | --- | ----------- | --------------- | ---------------- |
-// | **cardinal** | *cardinal number* | `<cardinal>7</cardinal>` | *siedem* |
-// | **signed** | *number with sign* | `<signed>-15</signed>` | *minus piętnaście* |
-// | **ordinal** | *ordinal number* | `<ordinal>1</ordinal>` | *pierwszy* |
-// | **fraction** | *fractional number* | `<fraction>3/4</fraction>` | *trzy czwarte* |
-// | **postal** | *postal code* | `<postal>30-020</postal>` | *trzydzieści zero dwadzieścia* |
-// | **time** | *time* | `<time>22:00</time>` | *dwudziesta druga* |
-// | **date** | *date* | `<date>12/05/2001</date>` | *dwunasty maja dwa tysiące jeden* |
-//
-// Note: when interpreting tags only nominal case is supported at the moment.
-//
-// You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. Currently supported option is only `sample_rate_hertz`, which is desired sampling frequency (in hertz) of synthesized audio.
+// You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. 
+// Service currently supports audio in two formats: sign 16 bit little-endian PCM wave or Ogg Opus. 
+// `sample_rate_hertz` can be set to change sampling rate of synthesized audio for wave PCM only.
 //
 // `SynthesizeRequest` can be sent to the service via gRPC insecure channel (that does not require authentication).
 //
-// `Synthesize` returns synthesized audio in `SynthesizeResponse` as a stream. When reading from the stream you have to check if `SynthesizeResponse` contains `error` field. If it does you can print its `code` and `description`. No `error` field in `SynthesizeResponse` means everything worked fine and its `audio` contains byte `content` that can be appended to received audio samples with `sample_rate_hertz` sampling frequency in hertz. When receiving `SynthesizeResponse` with `audio` you have to check if its `end_of_stream` flag is set to true. When it is set to true it means service has fnished synthesis and you can save your wave file with received synthesized audio content.
+// `Synthesize` returns synthesized audio in `SynthesizeResponse` as a stream. 
+// When reading from the stream you have to check if `SynthesizeResponse` contains `error` field. 
+// If it does you can print its `code` and `description`.
+// No `error` field in `SynthesizeResponse` means everything worked fine and its `audio` contains byte `content` 
+// that can be appended to received audio samples with `sample_rate_hertz` sampling frequency in hertz. 
+// When receiving `SynthesizeResponse` with `audio` you have to check if its `end_of_stream` flag is set to true. 
+// When it is set to true it means service has finished synthesis and you can save your wave file with received synthesized audio content.
 class TTS final {
  public:
   static constexpr char const* service_full_name() {
diff --git a/cpp/libtribune-client/tribune_tts.pb.cc b/cpp/libtribune-client/tribune_tts.pb.cc
index 591d5dc..91ecb71 100644
--- a/cpp/libtribune-client/tribune_tts.pb.cc
+++ b/cpp/libtribune-client/tribune_tts.pb.cc
@@ -51,7 +51,7 @@ namespace protobuf_tribune_5ftts_2eproto {
 namespace {
 
 ::google::protobuf::Metadata file_level_metadata[5];
-const ::google::protobuf::EnumDescriptor* file_level_enum_descriptors[1];
+const ::google::protobuf::EnumDescriptor* file_level_enum_descriptors[2];
 
 }  // namespace
 
@@ -87,6 +87,7 @@ const ::google::protobuf::uint32 TableStruct::offsets[] GOOGLE_ATTRIBUTE_SECTION
   ~0u,  // no _oneof_case_
   ~0u,  // no _weak_field_map_
   GOOGLE_PROTOBUF_GENERATED_MESSAGE_FIELD_OFFSET(SynthesizeConfig, sample_rate_hertz_),
+  GOOGLE_PROTOBUF_GENERATED_MESSAGE_FIELD_OFFSET(SynthesizeConfig, encoding_),
   ~0u,  // no _has_bits_
   GOOGLE_PROTOBUF_GENERATED_MESSAGE_FIELD_OFFSET(SynthesizeResponse, _internal_metadata_),
   ~0u,  // no _extensions_
@@ -113,9 +114,9 @@ const ::google::protobuf::uint32 TableStruct::offsets[] GOOGLE_ATTRIBUTE_SECTION
 static const ::google::protobuf::internal::MigrationSchema schemas[] GOOGLE_ATTRIBUTE_SECTION_VARIABLE(protodesc_cold) = {
   { 0, -1, sizeof(SynthesizeRequest)},
   { 7, -1, sizeof(SynthesizeConfig)},
-  { 13, -1, sizeof(SynthesizeResponse)},
-  { 20, -1, sizeof(AudioData)},
-  { 28, -1, sizeof(Error)},
+  { 14, -1, sizeof(SynthesizeResponse)},
+  { 21, -1, sizeof(AudioData)},
+  { 29, -1, sizeof(Error)},
 };
 
 static ::google::protobuf::Message const * const file_default_instances[] = {
@@ -181,22 +182,24 @@ void AddDescriptorsImpl() {
       "\n\021tribune_tts.proto\022\016techmo.tribune\"S\n\021S"
       "ynthesizeRequest\022\014\n\004text\030\001 \001(\t\0220\n\006config"
       "\030\002 \001(\0132 .techmo.tribune.SynthesizeConfig"
-      "\"-\n\020SynthesizeConfig\022\031\n\021sample_rate_hert"
-      "z\030\001 \001(\005\"d\n\022SynthesizeResponse\022(\n\005audio\030\001"
-      " \001(\0132\031.techmo.tribune.AudioData\022$\n\005error"
-      "\030\002 \001(\0132\025.techmo.tribune.Error\"N\n\tAudioDa"
-      "ta\022\031\n\021sample_rate_hertz\030\001 \001(\005\022\017\n\007content"
-      "\030\002 \001(\014\022\025\n\rend_of_stream\030\003 \001(\010\"E\n\005Error\022\'"
-      "\n\004code\030\001 \001(\0162\031.techmo.tribune.ErrorCode\022"
-      "\023\n\013description\030\002 \001(\t*_\n\tErrorCode\022\013\n\007UNK"
-      "NOWN\020\000\022\013\n\007LICENCE\020\001\022\026\n\022TEXT_NORMALIZATIO"
-      "N\020\002\022\021\n\rTRANSCRIPTION\020\003\022\r\n\tSYNTHESIS\020\0042\\\n"
-      "\003TTS\022U\n\nSynthesize\022!.techmo.tribune.Synt"
-      "hesizeRequest\032\".techmo.tribune.Synthesiz"
-      "eResponse0\001b\006proto3"
+      "\"^\n\020SynthesizeConfig\022\031\n\021sample_rate_hert"
+      "z\030\001 \001(\005\022/\n\010encoding\030\002 \001(\0162\035.techmo.tribu"
+      "ne.AudioEncoding\"d\n\022SynthesizeResponse\022("
+      "\n\005audio\030\001 \001(\0132\031.techmo.tribune.AudioData"
+      "\022$\n\005error\030\002 \001(\0132\025.techmo.tribune.Error\"N"
+      "\n\tAudioData\022\031\n\021sample_rate_hertz\030\001 \001(\005\022\017"
+      "\n\007content\030\002 \001(\014\022\025\n\rend_of_stream\030\003 \001(\010\"E"
+      "\n\005Error\022\'\n\004code\030\001 \001(\0162\031.techmo.tribune.E"
+      "rrorCode\022\023\n\013description\030\002 \001(\t*+\n\rAudioEn"
+      "coding\022\014\n\010LINEAR16\020\000\022\014\n\010OGG_OPUS\020\001*_\n\tEr"
+      "rorCode\022\013\n\007UNKNOWN\020\000\022\013\n\007LICENCE\020\001\022\026\n\022TEX"
+      "T_NORMALIZATION\020\002\022\021\n\rTRANSCRIPTION\020\003\022\r\n\t"
+      "SYNTHESIS\020\0042\\\n\003TTS\022U\n\nSynthesize\022!.techm"
+      "o.tribune.SynthesizeRequest\032\".techmo.tri"
+      "bune.SynthesizeResponse0\001b\006proto3"
   };
   ::google::protobuf::DescriptorPool::InternalAddGeneratedFile(
-      descriptor, 619);
+      descriptor, 713);
   ::google::protobuf::MessageFactory::InternalRegisterGeneratedFile(
     "tribune_tts.proto", &protobuf_RegisterTypes);
 }
@@ -215,10 +218,24 @@ struct StaticDescriptorInitializer {
 
 }  // namespace protobuf_tribune_5ftts_2eproto
 
-const ::google::protobuf::EnumDescriptor* ErrorCode_descriptor() {
+const ::google::protobuf::EnumDescriptor* AudioEncoding_descriptor() {
   protobuf_tribune_5ftts_2eproto::protobuf_AssignDescriptorsOnce();
   return protobuf_tribune_5ftts_2eproto::file_level_enum_descriptors[0];
 }
+bool AudioEncoding_IsValid(int value) {
+  switch (value) {
+    case 0:
+    case 1:
+      return true;
+    default:
+      return false;
+  }
+}
+
+const ::google::protobuf::EnumDescriptor* ErrorCode_descriptor() {
+  protobuf_tribune_5ftts_2eproto::protobuf_AssignDescriptorsOnce();
+  return protobuf_tribune_5ftts_2eproto::file_level_enum_descriptors[1];
+}
 bool ErrorCode_IsValid(int value) {
   switch (value) {
     case 0:
@@ -636,6 +653,7 @@ void SynthesizeRequest::set_allocated_config(::techmo::tribune::SynthesizeConfig
 
 #if !defined(_MSC_VER) || _MSC_VER >= 1900
 const int SynthesizeConfig::kSampleRateHertzFieldNumber;
+const int SynthesizeConfig::kEncodingFieldNumber;
 #endif  // !defined(_MSC_VER) || _MSC_VER >= 1900
 
 SynthesizeConfig::SynthesizeConfig()
@@ -651,12 +669,16 @@ SynthesizeConfig::SynthesizeConfig(const SynthesizeConfig& from)
       _internal_metadata_(NULL),
       _cached_size_(0) {
   _internal_metadata_.MergeFrom(from._internal_metadata_);
-  sample_rate_hertz_ = from.sample_rate_hertz_;
+  ::memcpy(&sample_rate_hertz_, &from.sample_rate_hertz_,
+    static_cast<size_t>(reinterpret_cast<char*>(&encoding_) -
+    reinterpret_cast<char*>(&sample_rate_hertz_)) + sizeof(encoding_));
   // @@protoc_insertion_point(copy_constructor:techmo.tribune.SynthesizeConfig)
 }
 
 void SynthesizeConfig::SharedCtor() {
-  sample_rate_hertz_ = 0;
+  ::memset(&sample_rate_hertz_, 0, static_cast<size_t>(
+      reinterpret_cast<char*>(&encoding_) -
+      reinterpret_cast<char*>(&sample_rate_hertz_)) + sizeof(encoding_));
   _cached_size_ = 0;
 }
 
@@ -697,7 +719,9 @@ void SynthesizeConfig::Clear() {
   // Prevent compiler warnings about cached_has_bits being unused
   (void) cached_has_bits;
 
-  sample_rate_hertz_ = 0;
+  ::memset(&sample_rate_hertz_, 0, static_cast<size_t>(
+      reinterpret_cast<char*>(&encoding_) -
+      reinterpret_cast<char*>(&sample_rate_hertz_)) + sizeof(encoding_));
   _internal_metadata_.Clear();
 }
 
@@ -725,6 +749,21 @@ bool SynthesizeConfig::MergePartialFromCodedStream(
         break;
       }
 
+      // .techmo.tribune.AudioEncoding encoding = 2;
+      case 2: {
+        if (static_cast< ::google::protobuf::uint8>(tag) ==
+            static_cast< ::google::protobuf::uint8>(16u /* 16 & 0xFF */)) {
+          int value;
+          DO_((::google::protobuf::internal::WireFormatLite::ReadPrimitive<
+                   int, ::google::protobuf::internal::WireFormatLite::TYPE_ENUM>(
+                 input, &value)));
+          set_encoding(static_cast< ::techmo::tribune::AudioEncoding >(value));
+        } else {
+          goto handle_unusual;
+        }
+        break;
+      }
+
       default: {
       handle_unusual:
         if (tag == 0) {
@@ -756,6 +795,12 @@ void SynthesizeConfig::SerializeWithCachedSizes(
     ::google::protobuf::internal::WireFormatLite::WriteInt32(1, this->sample_rate_hertz(), output);
   }
 
+  // .techmo.tribune.AudioEncoding encoding = 2;
+  if (this->encoding() != 0) {
+    ::google::protobuf::internal::WireFormatLite::WriteEnum(
+      2, this->encoding(), output);
+  }
+
   if ((_internal_metadata_.have_unknown_fields() &&  ::google::protobuf::internal::GetProto3PreserveUnknownsDefault())) {
     ::google::protobuf::internal::WireFormat::SerializeUnknownFields(
         (::google::protobuf::internal::GetProto3PreserveUnknownsDefault()   ? _internal_metadata_.unknown_fields()   : _internal_metadata_.default_instance()), output);
@@ -775,6 +820,12 @@ ::google::protobuf::uint8* SynthesizeConfig::InternalSerializeWithCachedSizesToA
     target = ::google::protobuf::internal::WireFormatLite::WriteInt32ToArray(1, this->sample_rate_hertz(), target);
   }
 
+  // .techmo.tribune.AudioEncoding encoding = 2;
+  if (this->encoding() != 0) {
+    target = ::google::protobuf::internal::WireFormatLite::WriteEnumToArray(
+      2, this->encoding(), target);
+  }
+
   if ((_internal_metadata_.have_unknown_fields() &&  ::google::protobuf::internal::GetProto3PreserveUnknownsDefault())) {
     target = ::google::protobuf::internal::WireFormat::SerializeUnknownFieldsToArray(
         (::google::protobuf::internal::GetProto3PreserveUnknownsDefault()   ? _internal_metadata_.unknown_fields()   : _internal_metadata_.default_instance()), target);
@@ -799,6 +850,12 @@ size_t SynthesizeConfig::ByteSizeLong() const {
         this->sample_rate_hertz());
   }
 
+  // .techmo.tribune.AudioEncoding encoding = 2;
+  if (this->encoding() != 0) {
+    total_size += 1 +
+      ::google::protobuf::internal::WireFormatLite::EnumSize(this->encoding());
+  }
+
   int cached_size = ::google::protobuf::internal::ToCachedSize(total_size);
   GOOGLE_SAFE_CONCURRENT_WRITES_BEGIN();
   _cached_size_ = cached_size;
@@ -831,6 +888,9 @@ void SynthesizeConfig::MergeFrom(const SynthesizeConfig& from) {
   if (from.sample_rate_hertz() != 0) {
     set_sample_rate_hertz(from.sample_rate_hertz());
   }
+  if (from.encoding() != 0) {
+    set_encoding(from.encoding());
+  }
 }
 
 void SynthesizeConfig::CopyFrom(const ::google::protobuf::Message& from) {
@@ -858,6 +918,7 @@ void SynthesizeConfig::Swap(SynthesizeConfig* other) {
 void SynthesizeConfig::InternalSwap(SynthesizeConfig* other) {
   using std::swap;
   swap(sample_rate_hertz_, other->sample_rate_hertz_);
+  swap(encoding_, other->encoding_);
   _internal_metadata_.Swap(&other->_internal_metadata_);
   swap(_cached_size_, other->_cached_size_);
 }
@@ -884,6 +945,20 @@ void SynthesizeConfig::set_sample_rate_hertz(::google::protobuf::int32 value) {
   // @@protoc_insertion_point(field_set:techmo.tribune.SynthesizeConfig.sample_rate_hertz)
 }
 
+// .techmo.tribune.AudioEncoding encoding = 2;
+void SynthesizeConfig::clear_encoding() {
+  encoding_ = 0;
+}
+::techmo::tribune::AudioEncoding SynthesizeConfig::encoding() const {
+  // @@protoc_insertion_point(field_get:techmo.tribune.SynthesizeConfig.encoding)
+  return static_cast< ::techmo::tribune::AudioEncoding >(encoding_);
+}
+void SynthesizeConfig::set_encoding(::techmo::tribune::AudioEncoding value) {
+  
+  encoding_ = value;
+  // @@protoc_insertion_point(field_set:techmo.tribune.SynthesizeConfig.encoding)
+}
+
 #endif  // PROTOBUF_INLINE_NOT_IN_HEADERS
 
 // ===================================================================
diff --git a/cpp/libtribune-client/tribune_tts.pb.h b/cpp/libtribune-client/tribune_tts.pb.h
index 2d6a509..2f75b1c 100644
--- a/cpp/libtribune-client/tribune_tts.pb.h
+++ b/cpp/libtribune-client/tribune_tts.pb.h
@@ -69,6 +69,27 @@ void AddDescriptors();
 void InitDefaults();
 }  // namespace protobuf_tribune_5ftts_2eproto
 
+enum AudioEncoding {
+  LINEAR16 = 0,
+  OGG_OPUS = 1,
+  AudioEncoding_INT_MIN_SENTINEL_DO_NOT_USE_ = ::google::protobuf::kint32min,
+  AudioEncoding_INT_MAX_SENTINEL_DO_NOT_USE_ = ::google::protobuf::kint32max
+};
+bool AudioEncoding_IsValid(int value);
+const AudioEncoding AudioEncoding_MIN = LINEAR16;
+const AudioEncoding AudioEncoding_MAX = OGG_OPUS;
+const int AudioEncoding_ARRAYSIZE = AudioEncoding_MAX + 1;
+
+const ::google::protobuf::EnumDescriptor* AudioEncoding_descriptor();
+inline const ::std::string& AudioEncoding_Name(AudioEncoding value) {
+  return ::google::protobuf::internal::NameOfEnum(
+    AudioEncoding_descriptor(), value);
+}
+inline bool AudioEncoding_Parse(
+    const ::std::string& name, AudioEncoding* value) {
+  return ::google::protobuf::internal::ParseNamedEnum<AudioEncoding>(
+    AudioEncoding_descriptor(), name, value);
+}
 enum ErrorCode {
   UNKNOWN = 0,
   LICENCE = 1,
@@ -297,11 +318,18 @@ class SynthesizeConfig : public ::google::protobuf::Message /* @@protoc_insertio
   ::google::protobuf::int32 sample_rate_hertz() const;
   void set_sample_rate_hertz(::google::protobuf::int32 value);
 
+  // .techmo.tribune.AudioEncoding encoding = 2;
+  void clear_encoding();
+  static const int kEncodingFieldNumber = 2;
+  ::techmo::tribune::AudioEncoding encoding() const;
+  void set_encoding(::techmo::tribune::AudioEncoding value);
+
   // @@protoc_insertion_point(class_scope:techmo.tribune.SynthesizeConfig)
  private:
 
   ::google::protobuf::internal::InternalMetadataWithArena _internal_metadata_;
   ::google::protobuf::int32 sample_rate_hertz_;
+  int encoding_;
   mutable int _cached_size_;
   friend struct protobuf_tribune_5ftts_2eproto::TableStruct;
 };
@@ -769,6 +797,20 @@ inline void SynthesizeConfig::set_sample_rate_hertz(::google::protobuf::int32 va
   // @@protoc_insertion_point(field_set:techmo.tribune.SynthesizeConfig.sample_rate_hertz)
 }
 
+// .techmo.tribune.AudioEncoding encoding = 2;
+inline void SynthesizeConfig::clear_encoding() {
+  encoding_ = 0;
+}
+inline ::techmo::tribune::AudioEncoding SynthesizeConfig::encoding() const {
+  // @@protoc_insertion_point(field_get:techmo.tribune.SynthesizeConfig.encoding)
+  return static_cast< ::techmo::tribune::AudioEncoding >(encoding_);
+}
+inline void SynthesizeConfig::set_encoding(::techmo::tribune::AudioEncoding value) {
+  
+  encoding_ = value;
+  // @@protoc_insertion_point(field_set:techmo.tribune.SynthesizeConfig.encoding)
+}
+
 // -------------------------------------------------------------------
 
 // SynthesizeResponse
@@ -1031,6 +1073,11 @@ inline void Error::set_allocated_description(::std::string* description) {
 namespace google {
 namespace protobuf {
 
+template <> struct is_proto_enum< ::techmo::tribune::AudioEncoding> : ::google::protobuf::internal::true_type {};
+template <>
+inline const EnumDescriptor* GetEnumDescriptor< ::techmo::tribune::AudioEncoding>() {
+  return ::techmo::tribune::AudioEncoding_descriptor();
+}
 template <> struct is_proto_enum< ::techmo::tribune::ErrorCode> : ::google::protobuf::internal::true_type {};
 template <>
 inline const EnumDescriptor* GetEnumDescriptor< ::techmo::tribune::ErrorCode>() {
diff --git a/cpp/tribune-client/main.cpp b/cpp/tribune-client/main.cpp
index f1d26e6..8d0ae92 100644
--- a/cpp/tribune-client/main.cpp
+++ b/cpp/tribune-client/main.cpp
@@ -1,4 +1,5 @@
 #include <iostream>
+#include <fstream>
 
 #include <boost/program_options.hpp>
 
@@ -26,7 +27,9 @@ po::options_description CreateOptionsDescription(void) {
              "how long the client is willing to wait for a reply from the server. "
              "If not specified, the service will set the deadline to a very large number.")
             ("sample-rate-hertz", po::value<unsigned int>()->default_value(0),
-             "Sample rate in Hz of synthesized audio. Set to 0 (default) to use voice's original sample rate.");
+             "Sample rate in Hz of synthesized audio. Set to 0 (default) to use voice's original sample rate.")
+            ("audio-encoding", po::value<std::string>()->default_value("LINEAR16"),
+             "Audio encoding, possible values are: LINEAR16, OGG_OPUS");
     return optionsDescription;
 }
 
@@ -58,12 +61,17 @@ int main(int argc, const char *const argv[]) {
         config.session_id = userOptions["session-id"].as<std::string>();
         config.grpc_timeout = userOptions["grpc-timeout"].as<int>();
         config.sample_rate_hertz = sample_rate_hertz;
-
+        const auto encoding = userOptions["audio-encoding"].as<std::string>();
+        const std::vector<std::string> allowedEncodings{ "LINEAR16", "OGG_OPUS" };
+        if(std::find(allowedEncodings.begin(), allowedEncodings.end(), encoding) == allowedEncodings.end()){
+            throw std::runtime_error("Unknown audio encoding: " + encoding);
+        }
+        techmo::tribune::AudioEncoding_Parse(encoding, &config.encoding);
         techmo::tribune::TribuneClient tribune_client{ userOptions["service-address"].as<std::string>() };
 
         const auto audio_data = tribune_client.Synthesize(config, userOptions["text"].as<std::string>());
 
-        WriteWaveFile(userOptions["out-path"].as<std::string>(), audio_data.sample_rate_hertz, audio_data.audio_bytes);
+        WriteFile(userOptions["out-path"].as<std::string>(), audio_data.sample_rate_hertz, config.encoding, audio_data.audio_bytes);
     }
     catch (const std::exception &e) {
         std::cerr << e.what() << std::endl;
diff --git a/cpp/tribune-client/wave-utils.cpp b/cpp/tribune-client/wave-utils.cpp
index 6e86aba..e444765 100644
--- a/cpp/tribune-client/wave-utils.cpp
+++ b/cpp/tribune-client/wave-utils.cpp
@@ -43,3 +43,16 @@ void WriteWaveFile(const std::string & wavePath, unsigned int sampleRate, const
     wave_file.write((char*)&audioBytes[0], audioBytes.size());
     wave_file.flush();
 }
+
+void WriteFile(const std::string & path, unsigned int sampleRate, techmo::tribune::AudioEncoding encoding, const std::string & audioBytes){
+    if(encoding == techmo::tribune::AudioEncoding::LINEAR16) {
+        WriteWaveFile(path, sampleRate, audioBytes);
+    }
+    else {
+        std::fstream file(path, std::ios::binary | std::ios::trunc | std::ios::out);
+        file.write(audioBytes.data(), audioBytes.size());
+        file.flush();
+    }
+
+
+}
diff --git a/cpp/tribune-client/wave-utils.h b/cpp/tribune-client/wave-utils.h
index 9163272..570b11c 100644
--- a/cpp/tribune-client/wave-utils.h
+++ b/cpp/tribune-client/wave-utils.h
@@ -3,6 +3,7 @@
 
 
 #include <string>
+#include "tribune_client.h"
 
 
 /// WAVE file header structure
@@ -34,8 +35,11 @@ struct WAV_DATA
 /// Reads at given path WAVE file header into WAV_HEADER struct and audio bytes into std::string.
 WAV_DATA ReadWaveFile(const std::string & wavePath);
 
-/// Writes at given path WAVE file with given sample rate [Hz] and audio bytes as std::string.
+/// Writes WAVE file at given path with given sample rate [Hz] and audio bytes as std::string.
 void WriteWaveFile(const std::string & wavePath, unsigned int sampleRate, const std::string & audioBytes);
 
+/// Writes file at given path depending on encoding.
+void WriteFile(const std::string & path, unsigned int sampleRate, techmo::tribune::AudioEncoding encoding, const std::string & audioBytes);
+
 
 #endif /* __WAVE_UTILS_H__ */
diff --git a/proto/tribune_tts.proto b/proto/tribune_tts.proto
index 0388efe..e2926ab 100755
--- a/proto/tribune_tts.proto
+++ b/proto/tribune_tts.proto
@@ -1,33 +1,29 @@
 // Techmo Tribune TTS API  
-// version: 1.1.0   
-// authors: Dawid Skurzok, Paweł Jaciów  
-// date:    2018-01-25  
+// version: 1.2.0   
+// authors: Dawid Skurzok, Paweł Jaciów, Jerzy Skrzypek
+// date:    2019-06-25  
 syntax = "proto3";
 
 package techmo.tribune;
 
-// Service that implements Techmo Text-To-Speech (TTS) API.
+// Techmo Text-To-Speech (TTS) API.
 //
 // Service's `Synthesize` method accepts `SynthesizeRequest` object which contains whole phrase to be synthesized.  
-// You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form. In that string you can use several special tags which can be interpreted. Tags have to be in from `<tag>something special</tag>` and can occur in any place in text. Currently interpreted tags are:
+// You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form.
 //
-// | Tag | Description | Example (input) | Example (output) |
-// | --- | ----------- | --------------- | ---------------- |
-// | **cardinal** | *cardinal number* | `<cardinal>7</cardinal>` | *siedem* |
-// | **signed** | *number with sign* | `<signed>-15</signed>` | *minus piętnaście* |
-// | **ordinal** | *ordinal number* | `<ordinal>1</ordinal>` | *pierwszy* |
-// | **fraction** | *fractional number* | `<fraction>3/4</fraction>` | *trzy czwarte* |
-// | **postal** | *postal code* | `<postal>30-020</postal>` | *trzydzieści zero dwadzieścia* |
-// | **time** | *time* | `<time>22:00</time>` | *dwudziesta druga* |
-// | **date** | *date* | `<date>12/05/2001</date>` | *dwunasty maja dwa tysiące jeden* |
-//
-// Note: when interpreting tags only nominal case is supported at the moment.
-//
-// You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. Currently supported option is only `sample_rate_hertz`, which is desired sampling frequency (in hertz) of synthesized audio.
+// You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. 
+// Service currently supports audio in two formats: sign 16 bit little-endian PCM wave or Ogg Opus. 
+// `sample_rate_hertz` can be set to change sampling rate of synthesized audio for wave PCM only.
 //
 // `SynthesizeRequest` can be sent to the service via gRPC insecure channel (that does not require authentication).
 //
-// `Synthesize` returns synthesized audio in `SynthesizeResponse` as a stream. When reading from the stream you have to check if `SynthesizeResponse` contains `error` field. If it does you can print its `code` and `description`. No `error` field in `SynthesizeResponse` means everything worked fine and its `audio` contains byte `content` that can be appended to received audio samples with `sample_rate_hertz` sampling frequency in hertz. When receiving `SynthesizeResponse` with `audio` you have to check if its `end_of_stream` flag is set to true. When it is set to true it means service has fnished synthesis and you can save your wave file with received synthesized audio content.
+// `Synthesize` returns synthesized audio in `SynthesizeResponse` as a stream. 
+// When reading from the stream you have to check if `SynthesizeResponse` contains `error` field. 
+// If it does you can print its `code` and `description`.
+// No `error` field in `SynthesizeResponse` means everything worked fine and its `audio` contains byte `content` 
+// that can be appended to received audio samples with `sample_rate_hertz` sampling frequency in hertz. 
+// When receiving `SynthesizeResponse` with `audio` you have to check if its `end_of_stream` flag is set to true. 
+// When it is set to true it means service has finished synthesis and you can save your wave file with received synthesized audio content.
 service TTS
 {
     // Returns audio signal with synthesized speech, given text and optional configuration.
@@ -45,11 +41,22 @@ message SynthesizeRequest
     SynthesizeConfig config = 2;
 }
 
+enum AudioEncoding {
+    // Signed 16 bit little endian PCM.
+    LINEAR16 = 0;
+    // Opus compressed audio in Ogg file container.
+    OGG_OPUS = 1;
+  }
+
 // Provides information to the synthesizer that specifies how to process the request.
 message SynthesizeConfig
 {
     // Desired sampling frequency in hertz of synthesized audio.
     int32 sample_rate_hertz = 1;
+
+    // Audio Format.
+    // Using OGG_OPUS will result in sample_rate_hertz being ignored.
+    AudioEncoding encoding = 2;
 }
 
 // `SynthesizeResponse` is the only message returned to the client by
diff --git a/python/CHANGELOG.md b/python/CHANGELOG.md
index a0fc361..8337b8e 100644
--- a/python/CHANGELOG.md
+++ b/python/CHANGELOG.md
@@ -1,5 +1,9 @@
 # Tribune TTS gRPC Python client Changelog
 
+## [1.3.0] - 2019-06-24
+### Added
+- `audio-encoding` option.
+
 ## [1.2.1] - 2019-04-19
 ### Fixed
 - Multithreading buffer access collision
diff --git a/python/VERSION.py b/python/VERSION.py
index 32a20ec..3258c7c 100644
--- a/python/VERSION.py
+++ b/python/VERSION.py
@@ -1 +1 @@
-TRIBUNE_CLIENT_VERSION = '1.2.1'
+TRIBUNE_CLIENT_VERSION = '1.3.0'
diff --git a/python/call_synthesize.py b/python/call_synthesize.py
index 11fc813..0aebfce 100644
--- a/python/call_synthesize.py
+++ b/python/call_synthesize.py
@@ -2,7 +2,7 @@
 import tribune_tts_pb2_grpc
 import grpc
 import os
-from wave_saver import WaveSaver
+from file_saver import SaverFactory
 
 
 def call_synthesize(args, text):
@@ -14,9 +14,11 @@ def call_synthesize(args, text):
     stub = tribune_tts_pb2_grpc.TTSStub(channel)
 
     # Synthesis request
-    config = tribune_tts_pb2.SynthesizeConfig(sample_rate_hertz=int(args.sample_rate))
+    config = tribune_tts_pb2.SynthesizeConfig(sample_rate_hertz=int(args.sample_rate), encoding=args.audio_encoding)
+    if config.encoding is tribune_tts_pb2.AudioEncoding.Value('OGG_OPUS') and config.sample_rate_hertz is not 0:
+        raise RuntimeError("Custom sample rate is not supported with Opus compression.")
     request = tribune_tts_pb2.SynthesizeRequest(text=text, config=config)
-    ws = WaveSaver()
+    ws = SaverFactory.get_saver(encoding=config.encoding)
 
     timeout=None
     if args.grpc_timeout > 0:
diff --git a/python/file_saver.py b/python/file_saver.py
new file mode 100644
index 0000000..ad0dd51
--- /dev/null
+++ b/python/file_saver.py
@@ -0,0 +1,35 @@
+import tribune_tts_pb2
+import tribune_tts_pb2_grpc
+import grpc
+import os
+from wave_saver import WaveSaver
+
+class SimpleFileSaver:
+    """Raw saver for TTS"""
+    buffer = None
+    _framerate = None
+
+    def __init__(self, sampling_frequency = None):
+        self.buffer = bytearray()
+        _framerate = None
+
+    def setFrameRate(self, sampling_frequency):
+        self._framerate = sampling_frequency
+
+    def append(self, audiodata):
+        self.buffer += audiodata
+
+    def clear(self):
+        self.buffer.clear()
+
+    def save(self, filename):
+        with open(filename, 'w+b') as f:
+            f.write(self.buffer)
+
+class SaverFactory:
+    @staticmethod
+    def get_saver(encoding):
+        if encoding == tribune_tts_pb2.AudioEncoding.Value('OGG_OPUS'):
+            return SimpleFileSaver()
+        else:
+            return WaveSaver()
diff --git a/python/tribune_client.py b/python/tribune_client.py
index 43cfc0c..003ad0a 100755
--- a/python/tribune_client.py
+++ b/python/tribune_client.py
@@ -22,7 +22,9 @@ def main():
                    help="Path to output wave file with synthesized audio content.", type=str)
     parser.add_argument("-f", "--sample_rate", dest="sample_rate", default=0,
                   help="Sample rate in Hz of synthesized audio. Set to 0 (default) to use voice's original sample rate.", type=int)
-    
+    parser.add_argument("--audio-encoding", dest="audio_encoding", default="LINEAR16",
+                        help="Audio encoding, possible values are: LINEAR16, OGG_OPUS", type=str)
+
     # Parse and validate options
     args = parser.parse_args()
 
diff --git a/python/tribune_tts_pb2.py b/python/tribune_tts_pb2.py
index f89dea2..4e77b7c 100644
--- a/python/tribune_tts_pb2.py
+++ b/python/tribune_tts_pb2.py
@@ -20,9 +20,32 @@
   name='tribune_tts.proto',
   package='techmo.tribune',
   syntax='proto3',
-  serialized_pb=_b('\n\x11tribune_tts.proto\x12\x0etechmo.tribune\"S\n\x11SynthesizeRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\x30\n\x06\x63onfig\x18\x02 \x01(\x0b\x32 .techmo.tribune.SynthesizeConfig\"-\n\x10SynthesizeConfig\x12\x19\n\x11sample_rate_hertz\x18\x01 \x01(\x05\"d\n\x12SynthesizeResponse\x12(\n\x05\x61udio\x18\x01 \x01(\x0b\x32\x19.techmo.tribune.AudioData\x12$\n\x05\x65rror\x18\x02 \x01(\x0b\x32\x15.techmo.tribune.Error\"N\n\tAudioData\x12\x19\n\x11sample_rate_hertz\x18\x01 \x01(\x05\x12\x0f\n\x07\x63ontent\x18\x02 \x01(\x0c\x12\x15\n\rend_of_stream\x18\x03 \x01(\x08\"E\n\x05\x45rror\x12\'\n\x04\x63ode\x18\x01 \x01(\x0e\x32\x19.techmo.tribune.ErrorCode\x12\x13\n\x0b\x64\x65scription\x18\x02 \x01(\t*_\n\tErrorCode\x12\x0b\n\x07UNKNOWN\x10\x00\x12\x0b\n\x07LICENCE\x10\x01\x12\x16\n\x12TEXT_NORMALIZATION\x10\x02\x12\x11\n\rTRANSCRIPTION\x10\x03\x12\r\n\tSYNTHESIS\x10\x04\x32\\\n\x03TTS\x12U\n\nSynthesize\x12!.techmo.tribune.SynthesizeRequest\x1a\".techmo.tribune.SynthesizeResponse0\x01\x62\x06proto3')
+  serialized_pb=_b('\n\x11tribune_tts.proto\x12\x0etechmo.tribune\"S\n\x11SynthesizeRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\x30\n\x06\x63onfig\x18\x02 \x01(\x0b\x32 .techmo.tribune.SynthesizeConfig\"^\n\x10SynthesizeConfig\x12\x19\n\x11sample_rate_hertz\x18\x01 \x01(\x05\x12/\n\x08\x65ncoding\x18\x02 \x01(\x0e\x32\x1d.techmo.tribune.AudioEncoding\"d\n\x12SynthesizeResponse\x12(\n\x05\x61udio\x18\x01 \x01(\x0b\x32\x19.techmo.tribune.AudioData\x12$\n\x05\x65rror\x18\x02 \x01(\x0b\x32\x15.techmo.tribune.Error\"N\n\tAudioData\x12\x19\n\x11sample_rate_hertz\x18\x01 \x01(\x05\x12\x0f\n\x07\x63ontent\x18\x02 \x01(\x0c\x12\x15\n\rend_of_stream\x18\x03 \x01(\x08\"E\n\x05\x45rror\x12\'\n\x04\x63ode\x18\x01 \x01(\x0e\x32\x19.techmo.tribune.ErrorCode\x12\x13\n\x0b\x64\x65scription\x18\x02 \x01(\t*+\n\rAudioEncoding\x12\x0c\n\x08LINEAR16\x10\x00\x12\x0c\n\x08OGG_OPUS\x10\x01*_\n\tErrorCode\x12\x0b\n\x07UNKNOWN\x10\x00\x12\x0b\n\x07LICENCE\x10\x01\x12\x16\n\x12TEXT_NORMALIZATION\x10\x02\x12\x11\n\rTRANSCRIPTION\x10\x03\x12\r\n\tSYNTHESIS\x10\x04\x32\\\n\x03TTS\x12U\n\nSynthesize\x12!.techmo.tribune.SynthesizeRequest\x1a\".techmo.tribune.SynthesizeResponse0\x01\x62\x06proto3')
 )
 
+_AUDIOENCODING = _descriptor.EnumDescriptor(
+  name='AudioEncoding',
+  full_name='techmo.tribune.AudioEncoding',
+  filename=None,
+  file=DESCRIPTOR,
+  values=[
+    _descriptor.EnumValueDescriptor(
+      name='LINEAR16', index=0, number=0,
+      options=None,
+      type=None),
+    _descriptor.EnumValueDescriptor(
+      name='OGG_OPUS', index=1, number=1,
+      options=None,
+      type=None),
+  ],
+  containing_type=None,
+  options=None,
+  serialized_start=471,
+  serialized_end=514,
+)
+_sym_db.RegisterEnumDescriptor(_AUDIOENCODING)
+
+AudioEncoding = enum_type_wrapper.EnumTypeWrapper(_AUDIOENCODING)
 _ERRORCODE = _descriptor.EnumDescriptor(
   name='ErrorCode',
   full_name='techmo.tribune.ErrorCode',
@@ -52,12 +75,14 @@
   ],
   containing_type=None,
   options=None,
-  serialized_start=422,
-  serialized_end=517,
+  serialized_start=516,
+  serialized_end=611,
 )
 _sym_db.RegisterEnumDescriptor(_ERRORCODE)
 
 ErrorCode = enum_type_wrapper.EnumTypeWrapper(_ERRORCODE)
+LINEAR16 = 0
+OGG_OPUS = 1
 UNKNOWN = 0
 LICENCE = 1
 TEXT_NORMALIZATION = 2
@@ -118,6 +143,13 @@
       message_type=None, enum_type=None, containing_type=None,
       is_extension=False, extension_scope=None,
       options=None),
+    _descriptor.FieldDescriptor(
+      name='encoding', full_name='techmo.tribune.SynthesizeConfig.encoding', index=1,
+      number=2, type=14, cpp_type=8, label=1,
+      has_default_value=False, default_value=0,
+      message_type=None, enum_type=None, containing_type=None,
+      is_extension=False, extension_scope=None,
+      options=None),
   ],
   extensions=[
   ],
@@ -131,7 +163,7 @@
   oneofs=[
   ],
   serialized_start=122,
-  serialized_end=167,
+  serialized_end=216,
 )
 
 
@@ -168,8 +200,8 @@
   extension_ranges=[],
   oneofs=[
   ],
-  serialized_start=169,
-  serialized_end=269,
+  serialized_start=218,
+  serialized_end=318,
 )
 
 
@@ -213,8 +245,8 @@
   extension_ranges=[],
   oneofs=[
   ],
-  serialized_start=271,
-  serialized_end=349,
+  serialized_start=320,
+  serialized_end=398,
 )
 
 
@@ -251,11 +283,12 @@
   extension_ranges=[],
   oneofs=[
   ],
-  serialized_start=351,
-  serialized_end=420,
+  serialized_start=400,
+  serialized_end=469,
 )
 
 _SYNTHESIZEREQUEST.fields_by_name['config'].message_type = _SYNTHESIZECONFIG
+_SYNTHESIZECONFIG.fields_by_name['encoding'].enum_type = _AUDIOENCODING
 _SYNTHESIZERESPONSE.fields_by_name['audio'].message_type = _AUDIODATA
 _SYNTHESIZERESPONSE.fields_by_name['error'].message_type = _ERROR
 _ERROR.fields_by_name['code'].enum_type = _ERRORCODE
@@ -264,6 +297,7 @@
 DESCRIPTOR.message_types_by_name['SynthesizeResponse'] = _SYNTHESIZERESPONSE
 DESCRIPTOR.message_types_by_name['AudioData'] = _AUDIODATA
 DESCRIPTOR.message_types_by_name['Error'] = _ERROR
+DESCRIPTOR.enum_types_by_name['AudioEncoding'] = _AUDIOENCODING
 DESCRIPTOR.enum_types_by_name['ErrorCode'] = _ERRORCODE
 _sym_db.RegisterFileDescriptor(DESCRIPTOR)
 
@@ -310,8 +344,8 @@
   file=DESCRIPTOR,
   index=0,
   options=None,
-  serialized_start=519,
-  serialized_end=611,
+  serialized_start=613,
+  serialized_end=705,
   methods=[
   _descriptor.MethodDescriptor(
     name='Synthesize',
diff --git a/python/tribune_tts_pb2_grpc.py b/python/tribune_tts_pb2_grpc.py
index 8341245..281a911 100644
--- a/python/tribune_tts_pb2_grpc.py
+++ b/python/tribune_tts_pb2_grpc.py
@@ -5,28 +5,24 @@
 
 
 class TTSStub(object):
-  """Service that implements Techmo Text-To-Speech (TTS) API.
+  """Techmo Text-To-Speech (TTS) API.
 
   Service's `Synthesize` method accepts `SynthesizeRequest` object which contains whole phrase to be synthesized.  
-  You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form. In that string you can use several special tags which can be interpreted. Tags have to be in from `<tag>something special</tag>` and can occur in any place in text. Currently interpreted tags are:
+  You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form.
 
-  | Tag | Description | Example (input) | Example (output) |
-  | --- | ----------- | --------------- | ---------------- |
-  | **cardinal** | *cardinal number* | `<cardinal>7</cardinal>` | *siedem* |
-  | **signed** | *number with sign* | `<signed>-15</signed>` | *minus piętnaście* |
-  | **ordinal** | *ordinal number* | `<ordinal>1</ordinal>` | *pierwszy* |
-  | **fraction** | *fractional number* | `<fraction>3/4</fraction>` | *trzy czwarte* |
-  | **postal** | *postal code* | `<postal>30-020</postal>` | *trzydzieści zero dwadzieścia* |
-  | **time** | *time* | `<time>22:00</time>` | *dwudziesta druga* |
-  | **date** | *date* | `<date>12/05/2001</date>` | *dwunasty maja dwa tysiące jeden* |
-
-  Note: when interpreting tags only nominal case is supported at the moment.
-
-  You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. Currently supported option is only `sample_rate_hertz`, which is desired sampling frequency (in hertz) of synthesized audio.
+  You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. 
+  Service currently supports audio in two formats: sign 16 bit little-endian PCM wave or Ogg Opus. 
+  `sample_rate_hertz` can be set to change sampling rate of synthesized audio for wave PCM only.
 
   `SynthesizeRequest` can be sent to the service via gRPC insecure channel (that does not require authentication).
 
-  `Synthesize` returns synthesized audio in `SynthesizeResponse` as a stream. When reading from the stream you have to check if `SynthesizeResponse` contains `error` field. If it does you can print its `code` and `description`. No `error` field in `SynthesizeResponse` means everything worked fine and its `audio` contains byte `content` that can be appended to received audio samples with `sample_rate_hertz` sampling frequency in hertz. When receiving `SynthesizeResponse` with `audio` you have to check if its `end_of_stream` flag is set to true. When it is set to true it means service has fnished synthesis and you can save your wave file with received synthesized audio content.
+  `Synthesize` returns synthesized audio in `SynthesizeResponse` as a stream. 
+  When reading from the stream you have to check if `SynthesizeResponse` contains `error` field. 
+  If it does you can print its `code` and `description`.
+  No `error` field in `SynthesizeResponse` means everything worked fine and its `audio` contains byte `content` 
+  that can be appended to received audio samples with `sample_rate_hertz` sampling frequency in hertz. 
+  When receiving `SynthesizeResponse` with `audio` you have to check if its `end_of_stream` flag is set to true. 
+  When it is set to true it means service has finished synthesis and you can save your wave file with received synthesized audio content.
   """
 
   def __init__(self, channel):
@@ -43,28 +39,24 @@ def __init__(self, channel):
 
 
 class TTSServicer(object):
-  """Service that implements Techmo Text-To-Speech (TTS) API.
+  """Techmo Text-To-Speech (TTS) API.
 
   Service's `Synthesize` method accepts `SynthesizeRequest` object which contains whole phrase to be synthesized.  
-  You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form. In that string you can use several special tags which can be interpreted. Tags have to be in from `<tag>something special</tag>` and can occur in any place in text. Currently interpreted tags are:
-
-  | Tag | Description | Example (input) | Example (output) |
-  | --- | ----------- | --------------- | ---------------- |
-  | **cardinal** | *cardinal number* | `<cardinal>7</cardinal>` | *siedem* |
-  | **signed** | *number with sign* | `<signed>-15</signed>` | *minus piętnaście* |
-  | **ordinal** | *ordinal number* | `<ordinal>1</ordinal>` | *pierwszy* |
-  | **fraction** | *fractional number* | `<fraction>3/4</fraction>` | *trzy czwarte* |
-  | **postal** | *postal code* | `<postal>30-020</postal>` | *trzydzieści zero dwadzieścia* |
-  | **time** | *time* | `<time>22:00</time>` | *dwudziesta druga* |
-  | **date** | *date* | `<date>12/05/2001</date>` | *dwunasty maja dwa tysiące jeden* |
-
-  Note: when interpreting tags only nominal case is supported at the moment.
+  You have to put the phrase as a string in `text` field of `SynthesizeRequest`. The string has to be in orthographic form.
 
-  You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. Currently supported option is only `sample_rate_hertz`, which is desired sampling frequency (in hertz) of synthesized audio.
+  You can set `SynthesizeConfig`'s fields to specify parameters of synthesis. 
+  Service currently supports audio in two formats: sign 16 bit little-endian PCM wave or Ogg Opus. 
+  `sample_rate_hertz` can be set to change sampling rate of synthesized audio for wave PCM only.
 
   `SynthesizeRequest` can be sent to the service via gRPC insecure channel (that does not require authentication).
 
-  `Synthesize` returns synthesized audio in `SynthesizeResponse` as a stream. When reading from the stream you have to check if `SynthesizeResponse` contains `error` field. If it does you can print its `code` and `description`. No `error` field in `SynthesizeResponse` means everything worked fine and its `audio` contains byte `content` that can be appended to received audio samples with `sample_rate_hertz` sampling frequency in hertz. When receiving `SynthesizeResponse` with `audio` you have to check if its `end_of_stream` flag is set to true. When it is set to true it means service has fnished synthesis and you can save your wave file with received synthesized audio content.
+  `Synthesize` returns synthesized audio in `SynthesizeResponse` as a stream. 
+  When reading from the stream you have to check if `SynthesizeResponse` contains `error` field. 
+  If it does you can print its `code` and `description`.
+  No `error` field in `SynthesizeResponse` means everything worked fine and its `audio` contains byte `content` 
+  that can be appended to received audio samples with `sample_rate_hertz` sampling frequency in hertz. 
+  When receiving `SynthesizeResponse` with `audio` you have to check if its `end_of_stream` flag is set to true. 
+  When it is set to true it means service has finished synthesis and you can save your wave file with received synthesized audio content.
   """
 
   def Synthesize(self, request, context):
diff --git a/python/wave_saver.py b/python/wave_saver.py
index db77c78..afdb126 100755
--- a/python/wave_saver.py
+++ b/python/wave_saver.py
@@ -31,7 +31,7 @@ def save(self, filename):
             params = (self._nchannels, self._sampwidth, self._framerate, len(self.buffer), 'NONE', 'not compressed')
             w.setparams(params)
             w.writeframes(self.buffer)
-    
+
     def load(self, filename):
         with wave.open(filename, 'r') as wr:
             self.buffer = wr.readframes(wr.getnframes())