Perceptual segmentation and component selection for sinusoidal representations of audio

Painter, Ted; Spanias, Andreas

doi:10.1109/tsa.2004.841050

Cited by 22 publications

(20 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Many different methods for solving this have been proposed, e.g. [21]- [28] all implement this in what seem to be different ways. Often, these methods rely heuristic rules taken from psychoacoustic experiments, while estimation theory, on the other hand, relies on statistical signal processing in finding model parameters.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

On perceptual distortion minimization and nonlinear least-squares frequency estimation

Christensen

Jensen

2006

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 99%

“…[26]. The methods of [27] and [28] are different methods yet-they rely on loudness and excitation pattern similarity criteria for sinusoidal component selection, respectively.…”

Section: Introductionmentioning

confidence: 99%

On perceptual distortion minimization and nonlinear least-squares frequency estimation

Christensen

Jensen

2006

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

“…Nilai periodisitas sinyal suara sangat berpengaruh pada persepsi penerima. Jika semua nilai puncak sinyal dipertahankan, maka periodisitas sinyal dapat tetap terjaga [17]. Nilai puncak yang masih tetap terjaga, akan memiliki tingkat periodisitas yang tinggi dan dapat dibuktikan dengan menggunakan deret Fourier.…”

Section: Pendahuluanunclassified

Model Sinusoida Secara Segmental Untuk Pengkodean Sinyal Suara

Setiawan¹,

Soegijoko²,

Sugihartono³

et al. 2010

MST

View full text Add to dashboard Cite

AbstrakSinyal yang periodik dapat didekomposisikan ke dalam bentuk sinusoida dengan menggunakan bantuan deret Fourier. Berdasarkan karakteristik sinyal suara yang demikian, maka dapat dilakukan pemodelan dengan mengacu pada bentuk sinusoida. Dengan menggunakan model sinusoida dapat dilakukan proses kuantisasi untuk mengkodekan sinyal suara pada laju yang rendah. Metode sinusoida telah banyak digunakan untuk mengkodekan sinyal suara. Dengan metode tersebut satu blok sinyal suara selebar 20 milidetik sampai dengan 30 milidetik dapat dikodekan dengan menggunakan koefisien deret Fourier. Metode baru yang diusulkan adalah kuantisasi dan rekonstruksi sinyal suara berdasarkan model sinusoida secara segmental. Segmen yang diambil adalah antara satu nilai puncak tertentu menuju ke nilai puncak berikutnya yang berlawanan, bukan berupa blok sinyal dengan panjang yang tetap seperti pada metode sinusoida yang sudah ada. Pengkode yang dirancang terdiri atas bagian enkoder dan dekoder. Enkoder berfungsi untuk mengkodekan sinyal suara pada laju variabel. Sinyal terkode selanjutnya dikirimkan ke penerima. Pada sisi penerima terdapat dekoder berfungsi untuk mengembalikan bentuk sinyal agar sesuai dengan asalnya dengan kualitas yang tidak jauh berbeda. Berdasarkan hasil percobaan diperoleh nilai rata-rata SNR segmental lebih dari 20 dB. AbstractSegmental Sinusoidal Model for Speech Signal Coding. Periodic signal can be decomposed by sinusoidal component with Fourier series. With this characteristic, it can be modeled referring by sinusoidal form. By the sinusoidal model, signal can be quantized in order to encode the speech signal at the lower rate. The recent sinusoidal method is implemented in speech coding. By using this method, a block of the speech signal with 20 ms to 30 ms width is coded based on Fourier series coefficients. The new method proposed is quantization and reconstruction of speech signal by the segmental sinusoidal model. A segment is defined as a block of the speech signal from certain peak to consecutive peak. The length of the segment is variable, instead of the fixed block like the recent sinusoidal method. Coder consists of the encoder and the decoder. Encoder works to code speech signal at variable rate. Then coded signal will be transmitted to receiver. On the receiver, coded signal will be reconstructed, so that the reconstruction signal has the near quality compared with the original signal. The experimental results show that the average of segmental SNR is more than 20 dB. : peak, period, quantization, segmental, sinusoidal Keywords PendahuluanJumlah kanal yang tersedia untuk komunikasi menjadi semakin terbatas seiring dengan makin pesatnya penggunaan kanal komunikasi. Kapasitas kanal yang terbatas mendorong untuk terus melakukan efisiensi di semua bagian pada sistem komunikasi. Komunikasi suara pada lebar pita 4 kHz pada format PCM 8-bit menyebabkan kanal harus menyediakan lebar pita 64 kbps untuk dapat mengirimkan informasi yang dapat ditangkap sesuai dengan aslinya. Pada perkembangannya, telah d...

show abstract

“…For example, parametric techniques based on mixed basis representations [12] and on Sinusoids + Transients + Noise (STN) models [3,4] have been successful in speech and audio synthesis. A number of important algorithms [2][3][4][5] have been proposed to estimate the amplitudes, frequencies and phases associated with the sinusoidal model. Some examples include peak picking in the short-time Fourier transform (STFT) domain [2], analysis-by-synthesis techniques [14], and matching pursuit decompositions [5].…”

Section: Introductionmentioning

confidence: 99%

“…For example, a strategy based on maximum signal to mask ratio (SMR) has been proposed for sinusoidal synthesis [1]. Additionally, excitation patterns [4] and loudness patterns [6] have also been employed for constrained sinusoidal representations. Peak-picking strategies based on maximum SMR or maximum SNR criteria focus on high-energy spectral regions, and therefore, can miss perceptually relevant sinusoids that are not located in these regions.…”

Section: Introductionmentioning

confidence: 99%

Low-complexity sinusoidal component selection using loudness patterns

Krishnamoorthi

Berisha

Spanias

et al. 2009

2009 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

Sinusoidal modeling of audio at low-bit rates involves selecting a limited number of parameters according to a quantitative or perceptual criterion. Most perceptual sinusoidal component selection strategies are computationally intensive and not suitable for real-time applications. In this paper, a computationally efficient sinusoidal selection algorithm based on a novel hybrid loudness estimation scheme is presented. The hybrid scheme first estimates efficiently the loudness of a multi-tone signal from the loudness patterns of its constituent sinusoidal components. Then it refines this estimate by performing a full evaluation of loudness but only in select critical bands. Experimental results show that the proposed technique maintains a low perceptual sinusoidal synthesis error at a much lower computational complexity.

show abstract

Perceptual segmentation and component selection for sinusoidal representations of audio

Cited by 22 publications

References 24 publications

On perceptual distortion minimization and nonlinear least-squares frequency estimation

On perceptual distortion minimization and nonlinear least-squares frequency estimation

Model Sinusoida Secara Segmental Untuk Pengkodean Sinyal Suara

Low-complexity sinusoidal component selection using loudness patterns

Contact Info

Product

Resources

About