OLD | NEW |
1 // Copyright (c) 2010 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2010 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 #include "chrome/browser/speech/audio_encoder.h" | 5 #include "chrome/browser/speech/audio_encoder.h" |
6 | 6 |
7 #include "base/basictypes.h" | 7 #include "base/basictypes.h" |
8 #include "base/logging.h" | 8 #include "base/logging.h" |
9 #include "base/scoped_ptr.h" | 9 #include "base/scoped_ptr.h" |
10 #include "base/stl_util-inl.h" | 10 #include "base/stl_util-inl.h" |
(...skipping 91 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
102 const char* const kContentTypeSpeex = "audio/x-speex-with-header-byte; rate="; | 102 const char* const kContentTypeSpeex = "audio/x-speex-with-header-byte; rate="; |
103 const int kSpeexEncodingQuality = 8; | 103 const int kSpeexEncodingQuality = 8; |
104 const int kMaxSpeexFrameLength = 110; // (44kbps rate sampled at 32kHz). | 104 const int kMaxSpeexFrameLength = 110; // (44kbps rate sampled at 32kHz). |
105 | 105 |
106 // Since the frame length gets written out as a byte in the encoded packet, | 106 // Since the frame length gets written out as a byte in the encoded packet, |
107 // make sure it is within the byte range. | 107 // make sure it is within the byte range. |
108 COMPILE_ASSERT(kMaxSpeexFrameLength <= 0xFF, invalidLength); | 108 COMPILE_ASSERT(kMaxSpeexFrameLength <= 0xFF, invalidLength); |
109 | 109 |
110 class SpeexEncoder : public speech_input::AudioEncoder { | 110 class SpeexEncoder : public speech_input::AudioEncoder { |
111 public: | 111 public: |
112 SpeexEncoder(int sampling_rate); | 112 explicit SpeexEncoder(int sampling_rate); |
| 113 virtual ~SpeexEncoder(); |
113 virtual void Encode(const short* samples, int num_samples); | 114 virtual void Encode(const short* samples, int num_samples); |
114 virtual void Flush() {} | 115 virtual void Flush() {} |
115 | 116 |
116 private: | 117 private: |
117 void* encoder_state_; | 118 void* encoder_state_; |
118 SpeexBits bits_; | 119 SpeexBits bits_; |
119 int samples_per_frame_; | 120 int samples_per_frame_; |
120 char encoded_frame_data_[kMaxSpeexFrameLength + 1]; // +1 for the frame size. | 121 char encoded_frame_data_[kMaxSpeexFrameLength + 1]; // +1 for the frame size. |
121 DISALLOW_COPY_AND_ASSIGN(SpeexEncoder); | 122 DISALLOW_COPY_AND_ASSIGN(SpeexEncoder); |
122 }; | 123 }; |
123 | 124 |
124 SpeexEncoder::SpeexEncoder(int sampling_rate) | 125 SpeexEncoder::SpeexEncoder(int sampling_rate) |
125 : AudioEncoder(std::string(kContentTypeSpeex) + | 126 : AudioEncoder(std::string(kContentTypeSpeex) + |
126 base::IntToString(sampling_rate)) { | 127 base::IntToString(sampling_rate)) { |
127 // speex_bits_init() does not initialize all of the |bits_| struct. | 128 // speex_bits_init() does not initialize all of the |bits_| struct. |
128 memset(&bits_, 0, sizeof(bits_)); | 129 memset(&bits_, 0, sizeof(bits_)); |
129 speex_bits_init(&bits_); | 130 speex_bits_init(&bits_); |
130 encoder_state_ = speex_encoder_init(&speex_wb_mode); | 131 encoder_state_ = speex_encoder_init(&speex_wb_mode); |
131 DCHECK(encoder_state_); | 132 DCHECK(encoder_state_); |
132 speex_encoder_ctl(encoder_state_, SPEEX_GET_FRAME_SIZE, &samples_per_frame_); | 133 speex_encoder_ctl(encoder_state_, SPEEX_GET_FRAME_SIZE, &samples_per_frame_); |
133 DCHECK(samples_per_frame_ > 0); | 134 DCHECK(samples_per_frame_ > 0); |
134 int quality = kSpeexEncodingQuality; | 135 int quality = kSpeexEncodingQuality; |
135 speex_encoder_ctl(encoder_state_, SPEEX_SET_QUALITY, &quality); | 136 speex_encoder_ctl(encoder_state_, SPEEX_SET_QUALITY, &quality); |
136 int vbr = 1; | 137 int vbr = 1; |
137 speex_encoder_ctl(encoder_state_, SPEEX_SET_VBR, &vbr); | 138 speex_encoder_ctl(encoder_state_, SPEEX_SET_VBR, &vbr); |
138 memset(encoded_frame_data_, 0, sizeof(encoded_frame_data_)); | 139 memset(encoded_frame_data_, 0, sizeof(encoded_frame_data_)); |
139 } | 140 } |
140 | 141 |
| 142 SpeexEncoder::~SpeexEncoder() { |
| 143 speex_bits_destroy(&bits_); |
| 144 speex_encoder_destroy(encoder_state_); |
| 145 } |
| 146 |
141 void SpeexEncoder::Encode(const short* samples, int num_samples) { | 147 void SpeexEncoder::Encode(const short* samples, int num_samples) { |
142 // Drop incomplete frames, typically those which come in when recording stops. | 148 // Drop incomplete frames, typically those which come in when recording stops. |
143 num_samples -= (num_samples % samples_per_frame_); | 149 num_samples -= (num_samples % samples_per_frame_); |
144 for (int i = 0; i < num_samples; i += samples_per_frame_) { | 150 for (int i = 0; i < num_samples; i += samples_per_frame_) { |
145 speex_bits_reset(&bits_); | 151 speex_bits_reset(&bits_); |
146 speex_encode_int(encoder_state_, const_cast<spx_int16_t*>(samples + i), | 152 speex_encode_int(encoder_state_, const_cast<spx_int16_t*>(samples + i), |
147 &bits_); | 153 &bits_); |
148 | 154 |
149 // Encode the frame and place the size of the frame as the first byte. This | 155 // Encode the frame and place the size of the frame as the first byte. This |
150 // is the packet format for MIME type x-speex-with-header-byte. | 156 // is the packet format for MIME type x-speex-with-header-byte. |
(...skipping 40 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
191 } | 197 } |
192 | 198 |
193 return true; | 199 return true; |
194 } | 200 } |
195 | 201 |
196 void AudioEncoder::AppendToBuffer(std::string* item) { | 202 void AudioEncoder::AppendToBuffer(std::string* item) { |
197 audio_buffers_.push_back(item); | 203 audio_buffers_.push_back(item); |
198 } | 204 } |
199 | 205 |
200 } // namespace speech_input | 206 } // namespace speech_input |
OLD | NEW |